Molecular Mechanisms of Oxygen Evolution Reactions for Artificial Photosynthesis

Abstract

Addressing the global environmental problem of water splitting to produce hydrogen fuel by solar energy is receiving so much attention. In water splitting, the essential problem to solve is the development of efficient catalysts for oxygen production. In this paper, having the prospect for a practical application of photocatalysts to artificial photosynthesis, molecular mechanisms in the current literature are briefly reviewed. At first, recent progress in the function of the Mn cluster at the natural photosystem II is briefly described. The kinds of devices in which oxygen evolution reaction (OER) catalysts are used were designated: water electrolyzers, photoelectrodes, and photocatalysts. Some methods for analyzing molecular mechanisms in OER catalysis, emphasized by the FTIR method, are shown briefly. After describing common OER mechanisms, the molecular mechanisms are discussed for TiO₂ and BiVO₄ photoelectrodes with our novel data, followed by presenting OER co-catalysts of IrO_2, RuO₂, NiO₂, and other metal oxides. Recent reports describing OER catalysts of perovskites, layered double hydroxides (LDH), metal–organic frameworks (MOF), single-atom catalysts, as well as metal complexes are reviewed. Finally, by comparing with natural photosystem, the required factors to improve the activity of the catalysts for artificial photosynthesis will be discussed.

1. Introduction

The photosynthetic system using solar energy has played a major role in the evolution of plants on the earth, which can oxidize water to generate molecular oxygen. Then, 2.4 billion years ago, the global environment was greatly changed for oxygen to occupy 1/5 of the atmosphere [1]. Before this change, living organisms used hydrogen sulfide as an electron source to live their life. In one billion years, living organisms evolved to utilize water as an electron source resulting in the oxygenic atmosphere. The organisms were further evolved by utilizing oxygen, and it was only about 0.2 million years ago that Homo sapiens were born. In recent years, the industrial revolution caused by human beings changed the global environment rapidly by using fossil fuels. Then, to replace fossil fuels with other chemical energy, the production of hydrogen with the aid of solar energy becomes indispensable. In addition, hydrogen will be used to industrially synthesize e-fuels by the reaction with CO₂.

In the generation of hydrogen by water splitting, molecular oxygen should be evolved at the same time. However, it is not so easy to artificially achieve photosynthetic oxygen generation by developing novel systems in a short period of time, which took one billion years in the natural world. The oxidation of water to produce oxygen, i.e., oxygen evolution reaction (OER), involves transfers of four electrons and four protons, while the reduction of water to produce hydrogen, i.e., hydrogen evolution reaction (HER), is a reaction of two electrons and two protons. Thus, the OER process is very difficult in comparison with the HER process. Actually, the overpotential of the present OER electrocatalysts is inevitably higher than that for the HER. In each oxidation step of the OER process, chemical species may appear as intermediates, which are unstable and then named reactive oxygen species (ROS). They are OH radicals, hydrogen peroxide, and superoxide radicals, which were actually detected in the photocatalytic reactions at the semiconductor surface [2]. To sustain the global environment, the reduction of CO₂ by solar light is also an important reaction. However, the CO₂ reduction catalysts are useful only when water is oxidized to generate oxygen at the same time [3]. Since oxygen evolution is the important key reaction, there are a lot of research papers describing the strategy to improve the OER efficiency with various kinds of methods, and the reaction mechanism was analyzed to increase productivity.

In the present paper, molecular mechanisms of water oxidation catalysis are briefly reviewed, with anticipating a practical application to artificial photosynthetic systems with photocatalysts [4]. Thus, this review is primarily focused on heterogeneous surface catalysis. At first, the function of the Mn cluster at the center of the photosystem II and the movement of protons to be released from water is briefly described. It is followed by the classification of catalysts for the OER employed, i.e., electrolyzer, photoelectrodes, and photocatalysts, in which the reaction mechanisms should have some difference. In the next section, the methods to investigate the molecular mechanism of OER processes are briefly described, where Fourier transfer infrared (FTIR) spectroscopy was mainly mentioned because it could detect the molecular structure in the OER process. Next is the widespread common mechanisms of OER, followed by the molecular mechanism of OER for each catalyst, which includes TiO₂, BiVO₄, and SrTiO₃ as semiconductors having the ability to direct excitation with photons. Especially for TiO₂ and BiVO₄, based on some unpublished data, some discussions were added to review the reported data. The following part of this report is devoted to the recent progress in the molecular mechanism of OER for various catalysts; IrO₂, RuO₂, transition metal oxides, CoOx, NiOx, FeOx, mixed metal oxides, perovskites, layered dihydroxides (LDHs), metal–organic frameworks (MOFs), metal complexes, single-atom catalysts (SAC), and so on. Finally, the characteristic difference between natural photosystem and present OER catalysts are pointed out, and the required factors to improve the activity of the catalysts for artificial photosynthesis will be discussed.

2. Oxygen Evolution Mechanism of Photosystem II (PS II)

In the photosynthesis of plants, across the thylakoid membrane, a series of electron transfers and proton transfers takes place simultaneously. That is, on the one side, two water molecules are oxidized to produce one molecular oxygen and four protons, and, on the other side of the membrane, high energy molecule ATP is produced from ADP in conjunction with the formation of an electron carrier, NADPH [5]. Oxygen is produced in light-driven water plastoquinone oxidoreductase enzymes, which are known as photosystem II (PS II). At the reaction center of PS II, photoexcitation energy is gathered with six pigments at the manganese–calcium–oxide-based Mn₄CaO₅ cluster in the reaction center, and then four electrons and four protons are transferred [6,7].

The radiation-damage-free structure of PS II was successfully analyzed with femtosecond X-ray free-electron lasers (XFEL). On the absorption of the second photon, the sixth O atom, O6, is coordinated at the position of 1.5Å from the fifth O atom, O5, of the Mn₄CaO₅ cluster, forming an O=O bond [8]. The O6 comes from H₂O, which was introduced by flipping the amine of the glutamic acid (Glu) residue [9].

Detailed time causes of electron transfer and proton transfer were analyzed by many spectroscopic methods, such as time-dependent fluorescence and transient absorption. The Mn₄CaO₅ cluster, at which molecular oxygen is generated, releases an electron to reduce the neighboring phenylalanine residue (Phe) [7]. Flash-induced FTIR difference spectroscopy revealed that the electron further moves to an iron quinone complex due to a small energy difference [10]. EPR measurements forecasted that, at the last quasi-stable state in four-step oxidation, the peroxo group, O-O bond is produced in the cluster, and immediately it turns into an O₂ molecule forming an O vacancy [11]. Density functional theory (DFT) calculation suggested that this O-O bond has the ionic state of [O₂]³⁻, and then antiferromagnetic interaction with the Mn ion stabilizes the O-O group to make a unique low-energy reaction path [12].

The application of time-resolved X-ray emission spectroscopy (XES) to each oxidation step revealed that oxidized tyrosine residue (Tyr) causes the fourth electron transfer, generating Mn^IV in the cluster, and all four Mn ions of the cluster became the valency of IV before the generation of the O-O bond [13]. Structural analysis was further developed [14], and the time-resolved structure after the photon absorption suggested the time cause of the position of the ligands and the water path [15]. The time-resolved IR spectra of microsecond resolution and computational chemistry showed the reaction time of electron transfer and proton transfer of four steps after each photon absorption as shown in Figure 1 [16].

In the photosynthesis cycle in Figure 1, the reaction process from S_3′ to S₄ was clarified recently as shown in Figure 2 [16]. (i) Immediately after the electron release from the CaMn₄O₅ cluster to Tyr, the proton coordinated to Glu moves to asparagine acid (Asp) residue by a tactical proton relay mechanism. (ii) O6 gets close to O5, and (iii) the O-O bond is formed.

A new prototype of a semiartificial system is reported by anchoring PS II on a polyethyleneimine-coated microporous carbon electrode. Practical applications may have difficulty with the low current density and durability [17]. Thus, the hybrid application may not be easy because PS II has been developed only to work in nature.

3. Devices for Oxygen Evolution

It has been reported since the 18th century that the application of electricity to the electrode meets with the dissociation of water to produce hydrogen and oxygen at a ratio of 2:1 [18]. This electrolysis reaction is practically used as a water electrolyzer to produce oxygen in the present day. About 50 years ago, it was found that a TiO₂ semiconductor electrode caused the formation of oxygen and hydrogen by irradiating light [19]. After the report of this photoelectrochemical method, a particulate photocatalytic reaction was tried to photo-split water into hydrogen and oxygen with many semiconducting materials [4]. As the developments in the water splitting methods, this section is divided into three subheadings: water electrolyzers, photoelectrochemical cells, and particulate photocatalysts.

3.1. Water Electrolyzer

Currently, 95% of total hydrogen is formed by steam reforming of fossil fuels [20]. On the other hand, 5% of hydrogen in the global market is produced by water electrolysis. The electricity capital expenses account for over 60% of H₂ content for renewable electricity if the electrolyzer can keep sustainable operation for 4000 h [21]. If the electricity could be served by solar panels, residual electricity caused by the monthly and daily variations of sunlight could be used to store the hydrogen energy. This system may actually contribute to the global environmental problems. From the viewpoint of the global transportation of energy, green hydrogen, as solar-driven chemical energy, should become indispensable in the near future, and then water electrolysis will become a widespread technology. Since the OER process is significantly difficult compared with the HER process as described above, the industrial scale of OER devices becomes important to produce green hydrogen.

As the water electrolyzer, there is an alkali system in which the anode and cathode are separated by a porous diaphragm [21], an acidic system with a proton exchange membrane (PEM) [22,23,24], and a solid oxide system with a high-temperature O²⁻ conductor [25]. Water electrolysis is the reverse reaction of hydrogen–oxygen fuel cells where chemical energy is changed to electric energy. Therefore, the equipment configuration of the apparatuses with PEM and solid electrolyte resemble the polymer electrolyte fuel cell (PEFC) and solid oxide fuel cell (SOFC), respectively. An alkaline water electrolyzer is conventionally used in industry, but the maximum current density is 400 mAcm⁻², and a long time is spent to start the operation. Then, the alkaline system is not favorable to use renewable energy with large fluctuation [25]. Since solid electrolytes need high temperatures, most of the research aims to acid water electrolysis [26]. The U.S. Department of Energy announced a goal to reach a lifetime of 80,000 h in 2050 [23].

To prepare a solar water splitting system, electrochemical water splitting systems can be easily integrated with photovoltaic systems. Actually, a solar to hydrogen (STH) conversion efficiency of 30% has been reported by connecting an available three-layer solar cell of InGaP/GaAs/GaInNAsS with an electrolyzer consisting of a Pt black cathode and an Ir black anode [27]. By integrating Pt and Ni on each surface of the GaInP/GainAs/Ge solar cell, the obtained photoelectrochemical cell showed 18% STH efficiency at 15 suns [28]. By using a commercially available IrRu-coated Ti anode connected to a Pt-coated Si photocathode by a perovskite solar cell, the STH efficiency was reported to be more than 17% [29]. Furthermore, by using a concentrated integrated photoelectrochemical device, a kilowatt-scale solar hydrogen production system is reported to be realized [30]. Though the higher STH efficiencies have been reported, enlargement of the devices and the cost for the practical operation may become problems.

The OER catalysts for water electrolysis have been developed mainly in alkaline and acidic conditions. Since photocatalytic water splitting is expected to take place mainly in a neutral solution, the OER catalysts for the electrolyzer may not be suitable for the photocatalytic system. However, water electrolysis in nearly neutral conditions has also been developed, which contains the OER catalysts used as co-catalysts of photocatalysts [31].

3.2. Photoelectrochemical Cell

Photoelectrochemical (PEC) cells based on an interfacial electric field formed at a semiconductor–electrolyte junction aim to achieve solar-driven water electrolysis at low cost and with low complexity [32]. This is the method to produce hydrogen by solar energy without using solar cells, which was discovered 50 years ago [33]. PEC devices have been investigated for many semiconducting materials. The water electrolyzers stated above are operated at a high current of more than 1 Acm⁻². On the other hand, since the uncondensed solar flux is less than 10 mAcm⁻², the operation condition of OER catalysts of PEC devices is different from that of electrolyzers. The materials that pose catalytic activity and durability under anodic polarization are limited to TiO₂, WO₃, BiVO₄, α-Fe₂O₃, β-NiOOH, and CuWO₄ [34,35,36,37].

A photoelectrochemical reaction with semiconductor electrodes was investigated in connection with the research in the reaction mechanism at the surface of visible-light-responsive photocatalysts; GaN [38], Ta₃N₅ [39,40], LaTiO₂N [41]. When BiVO₄ was used as a photoanode, a solar water splitting efficiency of 3.17% was reported [42]. To increase the OER efficiency, the loading of cobalt phosphate (CoPi) catalyst [43], and the application of high valence iron ion, Fe⁴⁺ [44], have been reported. The STH efficiency of the PEC device made of semiconductor photoelectrodes has not exceeded that of the above-mentioned combination device with a water electrolyzer and a photovoltaic cell. For solar-driven electron sources, OER at SrTiO₃ and BiVO₄ photoanodes were used not only for the generation of hydrogen but also for the reduction of CO₂ to generate formic acid [45] and carbon monoxide [46], respectively.

3.3. Particulate Photocatalysts

Particulate semiconductor photocatalysts seem to act as a small device of electrolyzer combined solar cell [47]. However, reactions on the electrode surface are essentially different from those on the particulate photocatalyst surface as discussed previously [48]. In the case of an electrochemical reaction, electrons are supplied continuously from the electric source through a lead wire. On the other hand, in the cases of photocatalysts and photoelectrodes, interfacial electron transfer reactions occur only at the absorption of photons. Thus, to proceed four-electron reaction, the reaction step of OER should wait until the absorption of the next photon. Moreover, in an electrochemical reaction, the electron transfer occurs at the electric field gradient formed at the surface of the electrode against the electrolyte in the solution, while in a photocatalytic reaction, the surface-trapped carrier generates the electric field gradient against the surface Fermi level [48]. Since the surface-trapped carrier may recombine with the carrier generated by the next absorption of a photon, the electron transfer reaction in photocatalysis should be rapid to overcome the rapid recombination reaction. Thus, the stability of the reaction intermediate, ROS in the case of OER, is most important to accomplish the process. The concept for effective photocatalysts is contrastive to the case of the PS II described above, where the reaction intermediate was not formed but the photon energy is accumulated in the valency of four Mn ions, and, on the third flash, O₂ is released immediately after the O-O bond formation.

There is an extensive review of water splitting with semiconductor photocatalysts [49,50]. The application of photocatalysis to the reduction of CO₂ has been compiled [3]. For practical testing of solar hydrogen generation, porous sheets of 100 m² made of photocatalyst powder were employed where the STH efficiency was 0.76% [51]. To increase the STH efficiency, p-n junction photocatalysts were proposed by combining two semiconductors to avoid recombination [52]. In a recent paper, by using Co₃O₄ as an OER co-catalyst on InGaN/GaN nanowire photocatalyst, STH of 9% in pure water, and 7% in seawater have been reported [53]. In a large-scale photocatalytic water-splitting system, an STH efficiency of 6.2% was described [53]. In comparison with the photovoltaic–electrolyzer combination systems, photocatalytic systems have some strong points that contain the scalability to enlarge for industrial manufacturers with abundant materials [49].

4. Methods for Investigating Mechanism

The molecular mechanism of OER at the atomic level is conveniently predicted by means of calculation chemistry. However, it should be supported by appropriate experimental measurements. In the field of photocatalysis, FTIR and electron paramagnetic resonance (EPR) spectroscopies under operation have been used to investigate the reaction mechanism. Operand/in situ measurements have recently been utilized for other spectroscopy such as X-ray absorption spectroscopy (XAS) and X-ray emission spectroscopy (XES), especially for electrocatalysts [54,55]. In this section, such analytical methods for catalyst surface, including operand X-ray diffraction (XRD) measurements, and electrochemical transmission microscopy (EC-TEM) will be described briefly.

4.1. Electrochemical Analysis

Electrochemistry is a conventional technique to investigate the electron transfer reaction at the solid surface. For kinetic analysis, linear sweep voltammetry (LSV), cyclic voltammetry (CV), and electrochemical impedance spectroscopy (EIS) are used. For the analysis of the catalytic activity, the initial rise and slope of the Tafel plot of LSV were used. Initial rise is represented by overpotential at 10 mA cm⁻², for example. The slope described in the unit of mV/dec is useful for microkinetic analysis, which can predict the rate-determining step in the four-step oxidation of OER [56,57,58]. In the cases of α-Fe₂O₃, BiVO₄, and TiO₂ electrodes, when the density of holes is small, the OER rate depends on the first order of the holes, indicating that the first step of OER is rate-determining. On the other hand, when the hole density is large, the OER rate depends on the third order, indicating the third oxidation step of rate-determining [59,60], which agrees with the prediction that the second oxidation step is the formation of relatively stable H₂O₂. Furthermore, in CV measurements, the isotope effect on the peak potential by using D₂O [61] could be used to investigate the mechanism of OER. These electrochemical measurements could not be applied to investigate semiconductor particulate photocatalysts, because the current at the particle surface cannot easily be measured.

4.2. Fourier Transform InfraRed (FTIR) Spectroscopy

FTIR (or IR) spectroscopy is a technique to measure the vibrational mode of bonds between atoms and, different from the Raman spectroscopy, is active only for the vibrational motions which can change dipole moment. Since the sensitivity is usually higher than the Raman spectroscopy, FTIR is the technique to obtain the key information for the molecular mechanism of catalysis [62]. Especially for the OER mechanism, FTIR could directly detect the adsorbed intermediate species, which is usually the ROS adsorbed on surface metal ions. To confine the detection area, the attenuated total reflectance (ATR) method could be used. The use of D₂O and H₂¹⁸O, which causes the isotope shift of the signal as predicted by the computer simulation, is helpful in assigning the signal. When linearly polarized IR light was used, the direction of the bond against the surface could be detected. Though FTIR spectroscopy is a traditional technique, a novel method with a synchrotron radiation source has been reported [63].

Table 1. FTIR peak position of the intermediate spaces in the OER processes.

Oxidation Step		/cm−1	Mode	Sample	[Reference] Year
1	Co(IV)=O	840	νMO	Co3O4	[64] 2014
	Ni-O·	908		Ni-NC	[65] 2020
	Ni-O·	916		Ni-Vac	[65] 2020
	Ni-O·	1170–1330		NiV-MOF	[66] 2021
	Mn(IV)=O	712–755		Mn complex	[67] 2001
	Mn(V)=O	970–981 1		Mn complex	[67] 2001
	Mn(VII)=O	912		NiFe-LDH	[68] 2018
	Fe(IV)=O	898		α-Fe2O3	[69] 2016
	Ti(IV)-O·	795		n-SrTiO3	[70] 2016
	Ir-O	784		Atomic Ir	[71] 2015
	Ir(VI)=O	946		CaIrO2	[63] 2022
2	Mn(IV)=O2	810		Mn complex	[72] 2021
	Li-OOH	1170		Li in DMSO	[73] 2017
	Ti-OOH	837	νOO	Ti-SiO2	[74] 2002
	Ti-OOH	838	νOO	TiO2	[75] 2003
	Ti-OOH	1120–1250	δOOH	TiO2	[75] 2003
	Ni-OOH	1060		Ni-Vac	[76] 2022
	Ni(IV)-OOD	1048		NiFe·MOF	[77] 2019
	Ni-OO-	850–1200		NiOOH	[78] 2019
	Cu-OO-	1180		Cu complex	[79] 2021
	Ru-OOH	986		RuO2	[80] 2021
	Ga-OOH	978		Ga2O3	[81] 2021
	Ir (VI)-OO-	870	νOO	CaIrO2	[63] 2022
	Ir(III)-OOH	830	νOO	IrO2	[82] 2011
	Ir-OOH	1055		Atomic Ir	[71] 2021
	Ir-OOH	1065		Ir/Co3O4	[83] 2023
	Au-OOH	1268		Au electrode	[84] 2005
	Co-OO-Co	833	νOO	Co3O4	[85] 2020
	Li-OO-Li	780, 830		Li in DMSO	[73] 2017
	Ti-OO-Ti	812		TiO2	[75] 2003
	Fe(VI)(=O)2	840, 856		NiFe-LDH	[86] 3018
	M-OO-M	1089		Ru/MnO2	[80] 2021
3	Co(III)-OO·(HO)	1013		Co3O4	[64] 2014
	Fe-OO·(HO)	1096		α-Fe2O3	[87] 2018
	Li-OO·	1127–1139		Li in DMSO	[73] 2017
	M-OO	1128		Ru/MnO2	[80] 2021
	Pt-OO·	1005–1016		Pt in Alkaline	[88] 2006
	Mn-OO	1080		Mn3O4	[89] 2019

lists the reported positions of FTIR signals for the OER intermediates adsorbed on several metal ions.

4.3. Raman Spectroscopy

Raman spectroscopy can be used to detect the vibration mode of the surface species similarly to the FTIR spectroscopy, and different from FTIR, in Raman spectroscopy, the signal of H₂O could not be observed [90]. Therefore, it is useful for applying to operand detection of the intermediates in OER. For example, the change in surface CoO to CoOOH in the OER reaction at the Co₃O₄ surface has been reported [91]. Since the signal intensity is generally weak, surface-enhanced Raman spectroscopy (SERS) with plasmonic metal particles [92,93] and tip-enhanced Raman scattering (TERS) with plasmonic SPM (scanning probe microscope) tips have been utilized [62].

4.4. Electron Paramagnetic Resonance (EPR) Spectroscopy

EPR spectroscopy is a unique method to directly detect paramagnetic chemical species with unpaired electrons that are caused at the surface of catalysts in both liquid and solid states [94,95]. Since the method is usually used to detect electron spins, it is called electron spin resonance (ESR) sometimes, or it should be called electron magnetic resonance (EMR) [96] by contrasting with nuclear magnetic resonance (NMR). Since metal atoms could have several spin states depending on the valency and coordination states, EPR is a useful technique to detect the situation around the active site. In addition to metal ions, it could also be used for detecting radical species derived from organic compounds and inorganic molecules.

4.5. Electrochemical Transmission Electron Microscope (EC-TEM)

Though the main part of TEM should be in an ultra-high vacuum, when the sample holder is arranged to be very thin and well shielded from the vacuum, operando TEM images could be obtained [97,98]. Furthermore, electron beam refraction patterns could be acquired [98]. Actually, a phase transition under the polarization was reported [98]. When EC-TEM was combined with electron energy loss spectroscopy (EELS) and energy dispersive spectroscopy (EDS), valence and distribution of positive ions under catalytic reactions could be investigated [99].

4.6. X-ray Photoelectron Spectroscopy (XPS)

XPS is used usually in a vacuum because electrons cannot easily move in the air, therefore it is difficult to measure the surface of the catalyst in a water environment. When the sample surface is separated from vacuum condition with a thin membrane, under normal pressure the XPS could be measured [100].

4.7. X-ray Absorption Spectroscopy (XAS)

XAS can be used to detect the surroundings of the X-ray-emitting metal. In the XAS method, X-ray absorption near-edge structure (XANES) gives the information of valence and electronic configuration, and extended X-ray absorption fine structure (EXAFS) is used to find out the interatomic distance, coordination number, and disorder of the coordination shell [101,102,103]. For electrocatalysts, operand XAS observation has been reported [102,104,105].

4.8. X-ray Emission Spectroscopy (XES)

XES is a technique that probes the radiative transitions from an occupied upper shell to an unoccupied or partially occupied core orbital. Since the excitation to the upper shell is required to cause the transition from the upper shell, XES provides complementary information of XAS. When the excitation is performed with XFEL or synchrotron radiation, time-resolved spectra can be obtained. Information about electronic structure, crystal field, and metal–ligand covalence could be obtained. When emission wavelength could be detected from soft X-ray to hard X-ray, measurable element spans from ligand elements, such as O, S, and P, to transition metals and actinides [61]. The apparatus used for XES is applicable for the measurements of resonant inelastic X-ray scattering (RIXS) and high energy resolution fluorescence detected XAS (HERFD-XAS) [61].

4.9. Mössbauer Spectroscopy

Mössbauer spectroscopy is a kind of resonance absorption technique in which the gamma ray of radioisotopes is used as the light source. Since the photon energy of gamma rays is very high, the sensitivity is high. The oxidation state, the spin state, and the symmetric property of ligands could be analyzed. However, nuclei that have a gamma ray emitting isotope are limited to Fe, Sn, Ir, and so on [106]. Recently, synchrotron radiation was used as the light source for the Mössbauer spectra of one of the actinide ¹⁶¹Dy [107]. The time-dependency of the ligand states [107] and the vibrational stets of the ligand of ⁵⁷Fe, were operando measured in CO₂ reduction reaction [108].

5. Common Mechanisms of OER Catalysis

Figure 3a,c,d show the catalytic mechanisms with metal oxides for OER reaction where four protons and four electrons are removed from two H₂O molecules to produce one O₂ molecule. The mechanisms are classified by whether the lattice oxygen is included (d) or not included (a,c).

In the case that the lattice oxygen is not included, when the reaction proceeds with only one metal ion as shown in Figure 3a, it is called the adsorbate evolution mechanism (AEM). While two adjacent metal ions contribute to the OER as shown in Figure 3c, it is called an oxide pair mechanism (OPM), which may be termed an oxide coupling mechanism (OCM), or interaction of two metal oxide entities (I2M) mechanism. On the other hand, when lattice oxygen participates in the mechanism, it is inclusively called lattice oxygen mechanism (LOM), and Figure 3d is an example of the possible LOM processes [61,109].

As shown in Figure 3b, on the assumption of the AEM process, H₂O is sequentially oxidized to *OH, *O, *OOH as the intermediates adsorbed on the surface, where * represents the adsorption site on the catalyst. The free energy can be calculated by computational method, usually DFT calculation is employed. For example, the free energy of each intermediate for OER catalysis of TiO₂, IrO₂, and WO₃ is shown in the literature [110]. Figure 3b schematically shows the free energy of each oxidation step in the AEM process. Since the oxidation energy of water is 1.23 eV × 4, the position in the free energy of *+H₂O and *+O₂ is fixed. The overpotential η is calculated from the maximum difference of the free energy ΔG_{n,max =} max{ΔG_n (n = 1, 2, 3, 4)} in the four energy steps as the difference from 1.23 eV, η = ΔG_n,max −1.23 eV. As the results of the calculation, it was shown that the free energies of *OH and *OOH are linearly correlated, meaning the energy difference between ΔG_*OH and ΔG_*OOH, that is ΔG₂ + ΔG₃ in Figure 3b, is constant to be 3.2 eV. Therefore, when ΔG_*O is located in the middle between ΔG_*OOH and ΔG_*OH the overpotential η becomes minimum. Therefore, when the expected overpotential η is plotted against ΔG₂ (= ΔG_*O − ΔG_*OH), η is expected to become a minimum of 0.37 at ΔG_{2 =} 1.6 eV. This is named scaling relationship, but it is adopted only on the AEM assumption [111].

In the reaction model of LOM, one of the O atoms of the oxygen molecule comes from lattice oxygen. Since there are several kinds of lattice oxygens, the reaction mechanism shown in Figure 3d is one of the possible LOM processes. In Figure 4, three other possible LOM processes are shown [112]. In the oxygen vacancy site mechanism (OVSM) shown in Figure 4a, a hole attacks the surface lattice oxygen, followed by a nucleophilic attack with a water molecule (WNA) causing the addition of the OH group at the lattice O atom. In the next hole, attacks cause the release of O₂ and a proton. The next oxidation step causes a surface lattice of OH from OH⁻ in water. With the fourth hole, surface OH releases a proton to return to the original surface O atom. In the single metal site mechanism (SMSM) shown in Figure 4b, when a hole comes into surface lattice oxygen, surface OH becomes surface OO by releasing a proton and making a surface O vacancy, and with the next hole, O₂ is released, and the surface O vacancy was filled by OH. At the next step, surface OH at the metal is recovered, and then at the next step, H of the surface lattice OH is removed to return to the original form. In the double metal site mechanism (DMSM) of Figure 4c, the protons of the adjacent two bridged OH groups are removed by two holes and the M-OO-M bond is formed at the surface. At the next step, O₂ is released, and the vacancy is covered with surface OH on the attack of the fourth hole.

In addition to the four LOM processes described above, other LOM paths can be drawn depending on the kind of metals and the surface crystalline lattice structure. Based on the assumption of the process, the oxidation steps of Gibbs free energy in the OER process can be theoretically calculated. It should be noted that removing the surface lattice oxygen in the LOM process suggests a high probability of the dissolution of metal ions.

6. Molecular Mechanism of OER

6.1. TiO₂

The history of TiO₂ for photo-induced water splitting and its molecular mechanism have been described in our previous report [113]. There are abundant reports in which molecular mechanisms of OER at the TiO₂ surface could be found. Research progress in the theoretical calculations for water splitting with TiO₂-based photocatalysts were also reported [114,115]. Most of the theoretical calculations were performed for rutile (110) and (001) surfaces. Recently, the rutile (100) surface was investigated for OER activity assuming the AEM process [116]. However, it has been shown experimentally that rutile powder consists of (110) and (011) or equivalent surfaces, and reduction occurs at the (110) surface while oxidation occurs at the (011) surface. Though the anatase (101) surface is also used in the theoretical calculation, many experimental observations have indicated that rutile TiO₂ is more active for OER than anatase TiO₂ [2,4]. This experimental conclusion could be explained by the authors as the surface Ti_5C-Ti_5C distance of rutile is shorter than that of anatase [117], and the recent theoretical calculation supported this explanation by comparing the anatase (101) surface with the rutile (110) surface [118].

In our previous report, a molecule Ti₅O₁₉H₁₆ modeling rutile (011) surface was used to investigate the OER process [113]. As the result of the total energy calculation, surface Ti_5C which can be coordinated by H₂O could make a hydrogen bonding with the facing bridged O atom. As shown in Figure 5, the first hole attacks the bridge O and then moves to the hydrogen bonding H₂O molecule to form Ti-O• at the facing Ti_5C which is the adsorbed OH radical. The second hole can attack Ti-O• or bridged O again to generate OH radical combined with Ti-O•, resulting in the formation of Ti-OOH species. This adsorbed hydroxyperoxo could be stabilized by forming a Ti-OO-Ti structure with another facing Ti_5C at the (011) surface. This calculation result suggested that the surface three Ti atoms concerning the OER process are not laid on a line but the third Ti of Ti_5C should be facing the bridged O [113].

Two decades ago, Nakamura et al. reported a molecular mechanism of water oxidation at rutile powder as shown in Figure 6 [119,120], and, in our previous review of the ROS formation [2], their reports were referred to in the discussion. In his mechanism, three-coordinated bridging O at a step or kink of the surface is attacked by a hole, simultaneously, WNA takes place to form Ti-O•, and then a pair of Ti-O• generate Ti-OO-Ti structure, and it becomes Ti-OOH [120]. Our prediction described above suggests that the third surface Ti atom may contribute to the following steps. The assignment of FTIR signals was added to this figure. Since the peak at 812 cm⁻¹ was not clearly shown in their in situ spectra [119], Ti-O• may directly become Ti-OOH by the second hole.

The molecular mechanism in Figure 6 was obtained by the observations with the ATR-FTIR for rutile powder and the STM and photoluminescence for rutile single crystals of (100) and (110) facets. The assignment of FTIR spectra was performed for photoinduced O₂ reduction under irradiation based on the isotope effect using D₂O and H₂¹⁸O [75]. For the signal of the Ti-OOH group, the isotope effect on the FTIR peaks is shown in

Table 2. FTIR peak position of hydroxyperoxo coordinated to TiO₂ and the isotope shifts.

	Mode	Ti-OOH	Ti-O18OH	Ti-18O18OH	Ti-OOD
Experimental a	δ(OOH)	1120–1250 cm−1			∆ = −220–−320 cm−1
	ν(OO)	838 cm−1	∆ = −19 cm−1	∆ = −45 cm−1	∆ = −67–−97 cm−1
DFT calculation b	δ(OOH)	1257 cm−1	∆ = −2 cm−1	∆ = −8 cm−1	∆ = −299 cm−1
	ν(OO)	881 cm−1	∆ = −23 cm−1	∆ = −49 cm−1	∆ = −22 cm−1

^a Experimental data were measured for rutile TiO₂ powder by Nakamura et al. J. Am. Chem. Soc. 2003 [75]. ^b DFT calculation by B3LYP/6-31G(d) method for a rutile (011) surface model molecule Ti₅O₁₉H₁₆ [113].

. The calculated signal position and the isotope effect are also listed in Table 2, which was performed for the above-mentioned (011) surface model. The signal assignment seems reasonable, and the isotope effect is well simulated. In the calculation, the signal intensity for bending mode δ(OOH) is larger than that of the stretching mode ν(OO), and the signal of the bending mode was clearly shown in the report [75]. However, in the in situ FTIR spectra for water oxidation, the signal at larger than 1100 cm⁻¹ in the spectrum was not shown [119]. The spectrum measured in H₂¹⁸O did not show the spreading near 1100 cm⁻¹. Since many chemical species have signals around 838 cm⁻¹, the observation of Ti-OOH in the experiment may have some problems. In our study, no peaks were observed around 850 cm⁻¹ [117]. They used Fe³⁺ ions for electron scavengers in the ATR-FTIR measurements, while we used oxygen as the electron scavenger. Based on the small shift observed for the 838 cm⁻¹ peak, they concluded the involvement of lattice oxygen in the OER process and deduced the molecular mechanism in Figure 6. Since the assignment of the observed peak has ambiguity, the OER process in Figure 6 may have some problems.

Having consensus on the assumption that the first oxidation takes place at bridged oxygen Ti-O-Ti, the reaction mechanism of OER was discussed [2]. Recently, Zhuang and Cheng [121] reported that at the rutile (110) surface, pKa of Ti_5COH₂ is larger than that of Ti_5COH-, and then the coordination structure Ti_5CO²⁻ exists stably in water and easily becomes Ti-O•. These results in the calculation seem to contradict the fact that the isoelectric point of rutile TiO₂ is around pH 6. The rutile (110) is not a minor surface in rutile powder as described above. It should be noted that theoretical calculations can bring the result without exception on the basis of the assumed model regardless of the experimental reality. Therefore, it should be careful to refer to the molecular mechanism derived only from the theoretical calculations. Thus, the calculation methods should be used only for the case justified with the experimental results.

6.2. BiVO₄

BiVO₄ was discovered first as the semiconductor photocatalyst to have the ability of water oxidation with visible light and the developed process was compiled in the report by Kudo et al. [122]. As the oxygen evolution photocatalysts, Mo-doped BiVO₄ embedded into an Au layer was employed in a Z-scheme photocatalytic system to exhibit 1.1% STH efficiency [123]. Aiming for utilization in photocatalytic oxidation, there are many reports that describe the effects of surface crystalline systems, morphologies, hetero junctions, and so on [124].

The electrochemical impedance technique was applied to investigate the micro kinetics at BiVO₄ photoelectrode and found the long-lived holes (0.1–1 s). Two kinds of recombination paths were suggested; one is the fast recombination with photogenerated electrons, and another is the recombination with BiVO₄ bulk electrons [125]. By measuring the lifetime of the photoluminescence, the recombination was found to occur in nanoseconds, which is probably caused by the multiphonon transition with deep-energy defects [126]. Sub-bandgap is formed by the V-deficiency to accelerate the recombination rate [127].

To understand the reaction mechanism, there are many computational studies in the literature. Walsh et al. concluded that the valence band (VB) of O_3p is coupled with Bi_6s to raise the VB maximum, while the conduction band (CB) consists of V_3d and O_2p and Bi_6p. Therefore, the adsorption of ROS is expected to take place at the Bi site [128]. On the other hand, in a later report for the (010) surface, O vacancy provides the V site as an active site. That is, the presence of the O vacancy increased the adsorption energy of H₂O, OH, O•, and the calculated free energy showed a decrease in the barrier for spontaneous charge transfer to electrolyte [129]. The doping of Co²⁺ was replaced with Bi³⁺, forming an O vacancy, and H₂O was replaced with the vacancy, which has been calculated to cause a decrease in the free energy by −0.28 eV [130]. At the (001) surface without vacancy, OH radical is easily formed, on the other hand at the (101) surface, where vacancies are generated by Mo/W doping, strong charge transfer to oxidation intermediates in OER takes place. This difference in the surface character causes charge transfer between (101) and (001) surfaces [131]. For the (010) surface, surface O vacancy is important to the adsorption of water in the catalytic activity [132]. The effect of O vacancy at the subsurface on the charge accumulation in the OER process was investigated. As a result, the O vacancy does not affect the photon acquisition or energy transportation in the crystal [133].

Twin-structured BiVO₄ was examined in the energy calculation [134]. The structure of the rate-determining step was considered with energy calculations in the OER process for four models (two AEM and two LOM). It was concluded that a larger amount of the twin structure causes a high OER activity [134].

Nikacevic et al. theoretically suggested some routes of OER at the BiVO₄ surface as shown in Figure 7. At the (001) surface, 97% of vacancies are coordinated with water. As shown in pathway A in this figure, Bi-OO-V is formed as an intermediate, and O₂ is evolved. On the other hand, only 0.05% of the O vacancy at the (011) surface was coordinated with water, but as shown in pathway B, through the Bi-OOH structure, O₂ is evolved as OER pathway B with a byproduct of H₂O₂ as the HPER pathway [135].

In the experimental research for the molecular mechanism of the BiVO₄ OER process, the surface interrogation scanning electrochemical microscopy technique was applied to W/Mo-doped BiVO₄ electrodes [136]. In this report, the generation of the OH radical was at the ratio of 6% of the absorbed photons in the OER process [136]. However, they detected the OH radical by the oxidation of IrCl₆²⁻ for electrochemical monitoring. For non-doped BiVO₄ electrodes, Nakabayashi et al. detected the OH radical by trapping it with a fluorescent reagent and reported that the yield of the OH radical was 0.06% of the photocurrent while almost 100% was used to generate molecular oxygen [137]. FTIR spectra under light irradiation were measured for the BiVO₄ photoanode [138]. The in situ FTIR spectra measured in H₂¹⁸O and D₂O are compiled in Figure 8. To analyze the isotope shift, DFT calculation with the B3LYP/LanL2DZ method in Gausian03W was performed for model molecule (OH)₄Bi-OOH, and the isotope shift is shown in

Table 3. FTIR peak position of hydroxyperoxo group coordinated to BiVO₄.

	Mode	Bi-OOH	Bi-O18OH	Bi-18O18OH	Bi-OOD
BiVO4 Experiment a	δ(OOH)	1233 cm−1		∆ = −8 cm−1	∆ = −313 cm−1
	ν(OO)	1035 cm−1		∆ = −50 cm−1	∆ = −45 cm−1
BiOOH DFT calc. b	δ(OOH)	1257 cm−1	∆ = −2 cm−1	∆ = −8 cm−1	∆ = −299 cm−1
	ν(OO)	881 cm−1	∆ = −23 cm−1	∆ = −50 cm−1	∆ = −22 cm−1

^a Taken from Ref [138] shown in Figure 8. ^b Calculated for (HO)₄BiOOH molecule.

. When the calculated isotope shift was checked with the measured FTIR spectra in the literature [138], the reported assignment of the peak in D₂O was found to be incorrect. The peak positions of δ(OOD) and ν(OO) should be alternated in D₂O. Then, the large isotope shift in H₂¹⁸O shows that both O in Bi-OOH originated from water and that the lattice oxygen of BiVO₄ was not contained in the OER process. Thus, experimental observations of the presence of Bi-OOH and the 100% yield of the O₂ show that the OER pathway B in Figure 7 is purely the molecular mechanism of OER at the BiVO₄ photoanode.

Though OER catalytic activity is observed for photoanode, to increase the separation of charges, BiVO₄ photoanode is sometimes used with the deposition of some OER catalysts such as Fe-OOH [139], CoPi [140], and Co₃O₄ [141].

6.3. SrTiO₃

Strontium titanate (SrTiO₃) is a well-documented photocatalyst in both one-step and two-step (Z-scheme) water-splitting systems as first reported in 1980 [142]. By using an ultrafast sub-band-gap probe with white light and 400 nm beam on SrTiO₃, one-electron intermediates were identified to be radicals located in Ti–O• (oxyl) and Ti–O•–Ti (bridge) groups with the initial radical formation time of 1.3 ps as shown in Figure 9 [143]. After photo-triggering the water oxidation reaction from the Nb-doped n-SrTiO₃ surface, the microsecond decay of the intermediates affirms transition-state theory through two distinct time constants. The reaction conditions can be adjusted to allow selection between the two pathways, one characterized by a labile intermediate facing the electrolyte (the oxyl), and the other by a lattice oxygen (the bridge) [144]. Picosecond optical spectra of the Ti-OH population on lightly doped SrTiO₃ are ordered by the surface hydroxylation. A Langmuir isotherm as a function of pH extracted an effective equilibrium constant relatable to the free energy difference of the first oxygen evolution reaction step [145].

Single crystalline SrTiO₃ photoanodes were irradiated with a focused laser to spatially define the degradation. By characterizing the degradation using optical spectroscopy and electron microscopy, the material dissolution constitutes an upper bound of 6% of the charge passed in a pH 13 electrolyte, while for pH 7, it reaches 23%; the pH dependence is anti-correlated with the ultrafast population of trapped charge. Since the dominant lattice dissolution reaction that occurred was thermodynamically concomitant with the OER, the reaction mechanism in Ref [144] was not supported, though the photon flux seems very high in this experiment [146].

6.4. Ga₂O₃

On Ga₂O₃-based photocatalysts, the overall water-splitting mechanism at a molecular level was suggested, based on the ATR-FTIR investigations combined with the mass spectroscopy (MS) analysis. Different from those observed in other semiconductor photocatalysts, a direct hydroxyl radical formation mechanism was suggested, where two adsorbed OH radicals became adsorbed H₂O₂ or adsorbed O₂²⁻, which caused the O₂ evolution with two other holes [81].

On different α-Ga₂O₃ surfaces, namely (001), (100), (110), and (012), water adsorption and activation were explored by means of DFT calculations. The dissociative water adsorption on all the studied low-index surfaces is thermodynamically favorable, and the most preferentially exposed surface is (012). The computation for the electrochemical model was used to investigate both the HER and the OER on α-Ga₂O₃ surfaces. The results indicate that the (100) and (110) surfaces are the most favorable ones for HER and OER, respectively, but they have low stability [147].

6.5. IrO₂ and RuO₂

Iridium oxide (IrOx) was used as the OER co-catalyst of the Y₂Ti₂O₅S₂ photocatalyst for visible light water splitting [148]. Ir-based catalysts are the catalysts of choice to date; nevertheless, their high price and scarcity have greatly hampered the widespread utilization of the PEM water electrolysis technique [149]. On the other hand, ruthenium (Ru), at higher earth abundance and lower price, possesses superior catalytic activity to Ir. However, Ru is prone to dissolution and results in low stability that cannot be applied in practical devices [149]. The surface stability and dissolution of three prominent electro(photo)catalysts for water splitting: RuO₂, IrO₂, and TiO₂ in the rutile phase was investigated by using a combination of ab initio steered molecular dynamics, enhanced sampling, and ab initio thermodynamics. A distinct site-specificity in the dissolution of the RuO₂(110) surface was identified, whereas no such surface site exists for the IrO₂(110) surface [150]. However, the mechanistic interplay between the OER and material degradation during water electrolysis is not yet well understood even for the most studied OER electrocatalysts such as IrO₂ and RuO₂ [151].

Anodically grown IrO_X catalyst films were studied using Raman spectroscopy. In addition to deuteration and ¹⁸O substitution experiments, theoretical models were also constructed using DFT to interpret the experimental data. The material was found to be composed of [IrO₆]_n edge-sharing polyhedra (with n ≥ 3) and characterized over a large potential range (0.0–1.8 V). Ir centers are connected to each other via μ-O type oxygen linkages that allow for the Ir centers to electronically couple to each other. Oxidation of Ir³⁺ to Ir⁴⁺ at 0.7–1.2 V within a μ-O linked polymeric geometry resulted in a blue coloration of the material at high potentials. Theoretical calculations indicated that the optical transition responsible for the color is essentially an Ir-to-Ir charge transfer transition [152].

For a catalyst system of Ir oxide nanocluster, a surface hydroperoxide, Ir-OOH, as an intermediate of OER has been detected by recording the FTIR spectra of the OO vibrational mode at 830 cm⁻¹. The detection was achieved upon oxidation of water under pulsed excitation of a visible light sensitizer [Ru(bpy)₃]²⁺ [82]. The OER mechanism of IrOx was investigated based on charge accumulation, where the valence change in Ir is more favorable than O–O bond formation. In situ evanescent wave spectroscopy revealed that an intermediate assignable to Ir⁵⁺ with oxygen ligands in opposite spin served as the precursor of OER regardless of pHs (2 to 12), as the generation of this species was not related to valence changes in Ir. The results confirmed that charge accumulation was not rate-limiting for OER on IrO_X, which is a key mechanistic difference between IrOx and less efficient 3d metal electrocatalysts [153]. Time-resolved operando spectroelectrochemistry was employed to investigate the redox-state kinetics of IrO_X electrocatalyst films for both water and hydrogen peroxide oxidation. Three different redox species involving Ir³⁺, Ir^3x+, Ir⁴⁺, and Ir^4y+ were identified spectroscopically. A first-order reaction mechanism was suggested for H₂O₂ oxidation driven by Ir⁴⁺ states, and a higher-order reaction mechanism involving the cooperative interaction of multiple Ir^4y+ states for water oxidation [154].

On calcined and uncalcined IrO_2, operando XAS spectroscopy was utilized to study the OER under different protocols. At the elevated OER potentials above 1.5 V, stronger Ir−Ir interactions were observed, which were more dominant in the calcined [155].

With first-principle calculations integrated with implicit solvation at constant potentials, the detailed atomistic reaction mechanism of OER was examined for the IrO₂(110) surface. The surface phase diagram was determined, and the possible reaction pathways including kinetic barriers, and computed reaction rates were explored based on the microkinetic models [156]. The classical mechanism at the IrO₂(110) surface was reconsidered. The OER follows a bi-nuclear mechanism with adjacent top surface oxygen atoms as fixed adsorption sites, whereas the Ir atoms underneath play an indirect role and maintain their saturated six-fold oxygen coordination at all stages of the reaction. The oxygen molecule is formed, via an Ir–OOOO–Ir transition state, by association of the outer oxygen atoms of two adjacent Ir–OO surface entities, leaving two intact Ir–O entities at the surface behind [157].

An IrO₂ nanoribbon of monoclinic phase, which is distinct from tetragonal rutile IrO₂, was provided by a molten alkali mechanochemical method. The intrinsic catalytic activity of IrO₂ nanoribbon was higher than that of rutile IrO₂ due to the low d band center of Ir in this special monoclinic phase structure as confirmed by DFT calculations [158]. Ultrasmall Pd@Ir core–shell nanoparticles (5 nm) with three atomic layers of iron–carbon nanotubes were constructed as an exceptional bifunctional electrocatalyst in acidic water splitting. Due to the core–shell structure, strain generated at heterointerfaces leads to an up-shifted d band center of Ir atoms contributing to a 62-fold better mass activity than commercial IrO₂. Moreover, electronic hybridization suppresses the electrochemical dissolution of Ir, resulting in the achievement of robust stability [159].

Ir_XRu_1−xO₂ (x = 1, 0.6, 0.3, and 0) was prepared by the hydrolysis synthesis, and a mechanistic study of the OER was reported. The polarization curves recorded at pHs of 0 to 3 could be well-fitted to a model consisting of a series of concerted electron–proton transfer reactions (a mononuclear mechanism). It was suggested that the third or fourth step is rate-determining for RuO₂ and IrO₂, respectively [160].

For single-crystal RuO₂(110) in acidic electrolytes, the surface structural changes as a function of potential were investigated by in situ surface X-ray scattering measurements with DFT calculations. The redox peaks at 0.7, 1.1, and 1.4 V vs. RHE could be attributed to the surface transitions associated with the successive deprotonation of *H₂O on the coordinately unsaturated Ru sites and hydrogen adsorbed to the bridging oxygen sites. At potentials relevant to the OER, an *OO species on the unsaturated Ru sites was detected, which was stabilized by a neighboring *OH group on the unsaturated Ru site or bridge site. A new OER pathway, where the deprotonation of the *OH group is used to stabilize *OO, was found to be rate-limiting [161]. For the RuO₂(110) surface, the DFT method considering a possible magnetic effect on the electronic configuration was applied to calculate the thermodynamic stability of possible O versus OH terminations and their effect on the free energies of the OER steps. The magnetic moment of RuO₂ supplies an important contribution to obtaining a low overpotential and to its insensitivity to the exact O versus OH coverage of RuO₂(110) surface [162].

The OER kinetics on RuO₂ rutile (110), (100), (101), and (111) orientations were experimentally investigated, finding (100) the most active. The potential involvement of lattice oxygen in the OER mechanism was assessed with online electrochemical mass spectrometry. No evidence of oxygen exchange on these oriented facets in acidic or basic electrolytes suggested that lattice oxygen is not exchanged in catalyzing OER on crystalline RuO₂ surfaces [163].

Rh doping for RuO₂ and surface oxygen vacancies to precisely regulate unconventional OER reaction paths via the Ru–O–Rh active sites have been reported. Quasi in situ/operando characterizations demonstrated the recurrence of reversible oxygen species under working potentials for enhanced activity and durability. It was theoretically revealed that Rh-RuO₂ passes through a reaction path of lattice oxygen-mediated mechanism, which is more optimal than the oxygen vacancy site mechanism. The synergistic interaction of defects and Ru–O–Rh active sites causes the *O formation with the rate-determining step, breaking the barrier limitation (*OOH) suggested by the traditional AEM process [164].

DFT calculations for RuO₂ demonstrated that the LOM can give rise to higher OER activity than the AEM at the active sites involving structural defects. Although the AEM is preferred for the perfect (110) and (211) surfaces, the formation of metal vacancies due to catalyst dissolution may lead to much lower OER overpotentials for the LOM. By screening several metal impurities in RuO₂, dopants such as Ni and Co can promote the LOM over the AEM even for perfectly structured surfaces [165]. Transition metal (TM)-doped rutile RuO₂ with different ratios of TM and Ru were discussed through DFT calculation with Hubbard U correction (+U). In low TM doping concentration, the evolved O₂ is generated through the AEM, and the OER activity is limited by the scaling relationship of OER intermediates. In higher TM doping concentrations, the evolved O₂ is generated through the LOM for Cu- or Ni-doped RuO₂. The distribution of Ru 4d and O 2p orbitals and the adsorption energy of H and O were found to be the major factors that affect the conversion of AEM into LOM [151].

Dispersing RuO₂ over defective TiO₂ enriched with oxygen vacancies (RuO₂/D-TiO₂) was reported with an electronic structure modulating strategy. A combination of (spectro-) electrochemistry and theoretical simulations revealed a continuous band structure at the interface between RuO₂ and defective TiO₂. Thus, a lowered energetic barrier for *OOH formation was found to be accountable for the largely enhanced acidic OER kinetics [166]. The effect of titanium substitution at different concentrations within nanoscale RuO₂, Ru_1−XTi_XO₂ (x = 0–50 atm%), on the structure, was reported for the OER activity and stability using combined experiments and theory [167]. For the MRuO_X solid solution (M = Ce⁴⁺, Sn⁴⁺, Ru⁴⁺, and Cr⁴⁺), the stability was customized by controlling the Ru charge. A scalable single-cell electrolyzer using a SnRuO_X anode and a polymer electrolyte membrane conveyed the ever-smallest degradation rate during a 1300 h operation at 1Acm⁻² [168]. For the sulfate-functionalized RuFeOx catalyst (S-RuFeOx) in a PEM of a water electrolyzer was investigated for OER activity and stability, because RuO₂ shows relatively poor stability. Coupled with the Fe cation doping, S-RuFeOx displayed a remarkable OER performance [169].

6.6. Perovskite as Electrocatalysts

Perovskite-type oxide nanocrystals (AxByOz), which possess distinct thermal stability, ionic conductivity, and electron mobility properties, have attracted increasing interest as efficient OER catalysts [170]. The electronic structure of perovskite-type nanocrystals plays a decisive role in electrocatalytic performance, the orbital filling, metal–oxygen hybridization, and electron correlations of perovskite-type oxide nanocrystals for high-performance OER catalysis were systematically investigated [170]. By using soft X-ray emission and absorption spectroscopies, perovskite OER catalysts were analyzed for the partial density of states on an absolute energy scale. Decreasing the solid-state charge-transfer energy of perovskite can change the mechanisms of the OER from electron-transfer-limited to proton–electron-coupled, and to proton-transfer-limited reactions [171]. The perovskite electrocatalysts with noble metals exhibited the smallest overpotential in various types of catalysts as shown in Figure 10 [172].

On various La_XSr_1−xCoO₃^−δ as OER catalysts, a general strategy was demonstrated for steering the two mechanisms, AEM and LOM. By delicately controlling the oxygen defect contents, the dominant OER mechanism can be arbitrarily transformed between AEM-LOM-AEM accompanied by a volcano-type activity variation trend. Experimental and computational evidence explicitly revealed that the phenomenon is due to the fact that the increased oxygen defects alter the lattice oxygen activity with a volcano-type trend and preserve the Co⁰ state for preferably the OER [173].

For NdNiO₃, the link between structural anisotropy and the OER catalytic activity was established by DFT calculation. The NdNiO₃ with (100), (110), and (111) orientations display similar oxidative states and metal–oxygen covalence characteristics, but distinct OER activities in experimental results were in the order of (100) > (110) > (111). DFT results confirm that film orientation is a critical determinant of the reaction mechanism. The OER on the (100)-surface favors proceeding via a LOM. In contrast, the reaction on (110)- and (111)-surfaces followed the AEM. The anisotropic oxygen vacancy formation energy and stability are strongly correlated to the reaction mechanism and performance [174]. On LaNiO₃ epitaxial thin films, electrochemical scanning tunneling microscopy (EC-STM) was used to directly observe structural dynamics during the OER. Based on the comparison of dynamic topographical changes in different compositions, reconstruction of surface morphology originating from the transition of the Ni species on the surface termination during the OER was proposed [175]. The change in surface topography was induced by Ni(OH)₂/NiOOH redox transformation by quantifying STM images [175].

On La_1−xNiO₃ perovskite electrocatalysts, direct O–O coupling promoted the OER activity at the interfacial active sites for decorated Ag nanoparticles. The theoretical calculation revealed that oxygen evolution via the dual-site mechanism with direct O–O coupling becomes more favorable than that via the conventional AEM. At x = 0.05, the electrocatalyst showed 20 times higher mass activity than that of the IrO₂ electrocatalyst, and the activity increased to 74 times after an accelerated durability test [176].

Ca_2−xIrO₄ nanocrystals exhibited a very high stability of about 62 times that of benchmark IrO₂. Lattice resolution images and surface-sensitive spectroscopies demonstrated the Ir-rich surface layer with high relative content of Ir⁵⁺ sites, which is responsible for the high activity and long-term stability. Combining operando IR spectroscopy with the XAS method, key intermediates of Ir⁶⁺ = O and Ir⁶⁺OO⁻ on Ir-based oxides electrocatalysts were observed, and they were stable even just from 1.3 V vs. RHE. DFT calculations indicated that the catalytic activity of Ca₂IrO₄ is enhanced remarkably after leaching the surface Ca ions because Ir = O and IrOO⁻ intermediates can be stabilized on positively charged active sites of the Ir-rich surface layer [63]. Layered perovskite Sr₂IrO₄ was chemically exfoliated into protonated colloidal nanosheets with an undamaged perovskite framework. This OER catalyst exhibited about 10 times higher activity than the IrO₂ catalyst film. As shown in Figure 11, DFT calculation indicated that electrons from the inner Ir atom to the surface were observed in IrO₂ (e) but not in the case of perovskite nanosheet (d), then the free energy of O* is moderate as shown in (a) compared with the case of IrO₂ (b). Thus, the structural hydroxyl groups on the surface of protonated nanosheets participate in the catalytic cycle [177].

6.7. Transition Metal (TM) Compounds

Multicomponent transition metal oxides and (oxy)hydroxides are the most promising OER catalysts due to their low cost, adjustable structure, high electrocatalytic activity, and outstanding durability. Co-, Ni-, and Fe-based OER catalysts have been considered to be potential candidates to replace noble metals, especially for electrocatalysts, due to their tunable 3d electron configuration and spin state, versatility in terms of crystal and electronic structures, as well as abundance in nature [23]. The latest advances in the rational design of the related OER electrocatalysts and the modulation of the electronic structure of active sites were comprehensively summarized, besides a brief overview of the mechanisms of OER and the theory and calculation criteria [178].

Wang et al. reviewed the fundamental understanding of the electronic structure of low-cost TM oxide-based catalysts for electrochemical OER, and its relationship with the catalytic activity and the reaction mechanism was discussed [179]. Feng et al. reviewed the relationship between TMs and OER catalyst activity, and then the mechanism of synthesis strategy in different types of TMs-based catalysts was summarized [180]. Guo et al. reviewed the state-of-the-art amorphous transition metal-based OER electrocatalysts, involving oxides, hydroxides, sulfides, phosphides, borides, and their composites, and then the practical application and theoretical modeling of the OER mechanisms in the OER were presented [181]. Though transition metal phosphides often exhibit excellent HER activity, the OER catalytic performance is not outstanding. Huang et al. reviewed the strategies for preparing highly active OER catalysts of transition metal phosphides [182].

The early transition metals (Ti, V, and Cr) can form very stable M=O units, while the late transition metals (Ni and Cu) can only theoretically form unstable M=O structures. On the other hand, for Mn, Fe, and Co, the metal-oxo motif switches between two valence tautomers in the form of Mn⁺¹ = O²⁻ and Mn–O•⁻. The former with an electrophilic oxygen atom can proceed via the acid–base WNA pathway to form the O–O bond, whereas the latter favors the oxygen radical coupling pathway for O–O bond formation [183].

6.7.1. CoOx

CoOOH was selected as the OER co-catalyst of aluminum-doped strontium titanate (SrTiO:Al) photocatalyst to attain almost unity in the internal quantum efficiency of UV-induced water splitting with Rh/Cr HER co-catalyst [184]. The recent progress of Co₃O₄-based electrocatalytic materials for the acidic OER was presented with particular reference to the catalytic mechanism and guidelines for the design principles from both experimental and theoretical perspectives [185]. Afterward, emerging strategies were outlined to improve the catalytic performance of Co₃O₄-based acidic OER catalysts, including phase engineering, component regulation with doping, composite with carbon-based materials, and multi-phase hybridization [185].

For the application of Co oxides to photocatalysts, operando XPS measurements were performed. The catalyst undergoes chemical–structural transformations as a function of the applied anodic potential, with complete conversion of the Co(OH)₂ and partial conversion of the spinel Co₃O₄ phases to cobalt oxyhydroxide, CoO(OH), under precatalytic electrochemical conditions. This interpretation revealed that the presence of Co(OH)₂ enhances catalytic activity by promoting transformations to CoO(OH) [186]. To study the mechanism of OER on CoO(OH), operando X-ray absorption and Raman spectroscopy revealed that a Co(IV) species, CoO₂, is the dominating resting state of the catalyst. Oxygen isotope exchange experiments showed that a cobalt superoxide species is an active intermediate in the OER. This intermediate is formed concurrently with the oxidation of CoO(OH) to CoO₂. Combining spectroscopic and electrokinetic data, the rate-determining step of the OER was identified as the release of dioxygen from the superoxide intermediate [92].

By using a water-in-salt electrolyte, the water activity was systematically tuned and the mechanism as a function of applied potentials in water electrolysis was probed. The mechanism is sensitive to the applied potential. The Co-OO-Co bond forms via an intramolecular oxygen coupling mechanism at low potentials, whereas it proceeds through a WNA mechanism by forming Co-OOH at high potentials [187].

The morphology-dependent analysis for well-defined crystalline CoO(OH) revealed that the active sites are exclusively located at lateral facets rather than basal facets. Theoretical calculations show that the coordinately unsaturated Co sites of lateral facets upshift the O 2p-band center closer to the Fermi level, thereby enhancing the covalency of Co-O bonds to yield the reactivity [188]. The sequential oxidation kinetics with Co₃O₄ nanoparticles involving multi-active sites for water oxidation in the OER catalytic cycle were resolved by applying quasi-operando transient absorption spectroscopy to a typical photosensitization with Ru-dye and sacrificial electron donor. The Co^IV intermediate distribution plays a determining role in OER activity and results in the slow overall OER kinetics [189]. The redox process between Co^III and Co^IV species does not follow a proton-coupled electron transfer mechanism that is thought to be common prior to the OER, but it involves a proton-decoupled electron transfer, clarified by isotope labeling experiments and in situ electrostatic modulation [190].

An oxygen vacancy (Vo)-rich environment facilitates the reconstruction of Co₃O₄ to the Co(OH)₂ intermediate with proton vacancies (Co^IIO_x(OH)_y), which is favorable for the formation of the active species of CoO(OH). Correlative operando Raman spectra characterizations and electrokinetic analyses indicated that a moderate Vo density can switch the O–O bond formation pathway, from a WNA to an INA pathway, which is more kinetically favorable for water oxidation [191]. As shown in Figure 12b with O vacancy, at step 3, three protons and one electron are removed to form Co-O(OH). At step 4, Co^III sites of Co-O(OH) are oxidized to Co^IV which can be deprotonated (step 8) by hole attack oxo ligand Co^IV = O forms a Co-O-O triangle (step 9), and then becomes Co^II-OO• (step 10). At the next oxidation (step 10), O₂ is released and Co^II back to Co^III with the coordination of water. The I2M process was excluded based on the experimental results of using the H₂¹⁸O isotope [191].

Amorphous CoO(OH) layer architecture was loaded onto the surface of TiO₂. Tafel analysis, EIS, and CV methods showed that the carrier transfer barrier within the electrode and the transition of Co^IIIO(OH) to Co^IVO(OH) have the dominating effects on the photoelectrochemical performance. Theoretical calculation revealed that the interface between the CoO(OH) and TiO₂ improves the electronic transfer ability among Co sites [192]. Amorphous CoO(OH) layers are electrochemically synthesized on the surface of various cobalt sulfides CoS_α and found to decrease the intermolecular energy gap. The decrease in the energy gap accelerates the formation of OER-active high-valent Co^IV species [193].

Co₃O₄ nanocrystals anchored on carbon nitride nanofiber (CNF) were prepared and found that the OER activity under visible light increased by 124 times, where heterogeneous kinetics is improved based on a synergistic effect between its binary components for charge separation and the (222) facet exposure of Co₃O₄ nanocrystals. DFT calculations revealed that Vo at (222) facet leads to a reduction of the bandgap of the nanocrystals [194].

The water oxidation with Co cubane cluster, Co₄O₄(OAc)₄(py)₄, as the catalysis was examined by time-resolved rapid scan ATR FTIR spectroscopy, the μ-peroxido structure, Co-OO-Co, was established as the intermediate. Where the one-electron oxidized cubane was the sole source of charge that was driven either in alkaline solution by a visible light sensitizer or in hydroxide (OH⁻) containing acetonitrile solution [85].

The simulations on the OER mechanism were performed and, in addition, the influence of Fe substitution was examined. Co^IV in the pristine CoO(OH) promotes the efficient formation of an active O radical intermediate followed by intramolecular O–O coupling. In the case of Fe substitution, the early oxidation of Fe^III to Fe^IV promotes the electrophilic character in the reaction center, reducing the proton affinity of the surface-bound hydroxyl moieties [195].

6.7.2. NiOx

For the nascent ultra-small nickel oxyhydroxide (NiOOH) particles (<3 nm), the thermodynamics of Ni dissolution was calculated by using first-principles theory at a near-neutral pH range, and the mechanism of OER on the γ-NiOOH surface was clarified. It was concluded that (i) ∼4% Ni cations on the surface of γ-NiOOH dissolve at pH = 7 and 1.73 V vs. RHE; (ii) on the pristine γ-NiOOH surface, OER proceeds via the “lattice peroxide” mechanism (*H₂O → *OH → *O–O_lattH* → O–O_latt → O₂) with an overpotential of 0.70 V; (iii) in the presence of Ni cationic vacancies, OER proceeds via the “hydroperoxide” mechanism (*OH + *H₂O → *2OH → *OOH → O₂) with an overpotential of 0.40 V [196].

For NiOOH-based materials, light-triggered reversible geometric conversion between octahedron (NiO₆) and square planar (NiO₄) was proposed. The unit cell was undergone to achieve electronic states with alternative metal and oxygen characters throughout the oxygen evolution process. Utilizing this electron transfer pathway can bypass the potential limiting steps, that is, O–O bonding in the AEM and deprotonation in the LOM. As a result, the electrocatalysts that operate through this route showed superior activity compared with previously reported electrocatalysts [197,198].

By incorporating Fe and V into Ni(OH)₂ lattices, OER activity was improved. X-ray photoelectron/absorption spectroscopies revealed the synergistic interaction between Fe/V dopants and Ni in the host matrix, which subtly modulates local coordination environments and electronic structures of the Fe/V/Ni cations. Further, in situ XAS analyses manifested contraction of metal–oxygen bond lengths in the activated catalyst, with a short V–O bond distance. DFT calculations indicated that the V site of the Fe/V co-doped NiOOH gave near-optimal binding energies of OER intermediates and had lower overpotential compared with Ni and Fe sites [199]. A series of Mn-, Co-, Fe-, and Zn-doped nickel oxides were investigated by using operando UV−vis spectroscopy coupled with time-resolved stepped potential spectroelectrochemistry. The Ni²⁺/Ni³⁺ redox peak potential was found to shift anodically from Mn- < Co- < Fe- < Zn-doped samples, suggesting a decrease in oxygen binding energetics from Mn- to Zn-doped samples. The OER kinetics had a second-order dependence on the density of these oxidized species, suggesting a chemical rate-determining step involving the coupling of two oxo species. The intrinsic turnover frequency per oxidized species exhibits a volcano trend with the binding energy of oxygen on the Ni site, having a maximum activity for the Fe-doped sample as shown in Figure 13. For Ni centers that bind oxygen too strongly (Mn- and Co-doped oxides), OER kinetics is limited by O–O coupling and oxygen desorption, while for Ni centers that bind oxygen too weakly (Zn-doped oxides), OER kinetics is limited by the formation of oxo groups [200].

Oxygen vacancy-enriched porous NiO/ln₂O₃ nanofibers (Vo–NiO/ln₂O₃@NFs) were fabricated for efficient OER electrocatalysis. Abundant Vo modulated the electronic configuration of the catalyst for altering the adsorption of intermediates to reduce the OER overpotential and promote *O formation, upshifting the d band center of metal centers near the Fermi level, and also increasing the electrical conductivity and enhancing the OER reaction kinetics simultaneously. In situ Raman spectra suggested that the Vo can render the NiO/ln₂O₃ more easily reconstructible on the surface during the OER course [201].

DFT +U calculations revealed that Ir-doping of a β-NiOOH(001) surface enhanced the electric conductivity while also activating an oxygen site involving three Ni atoms to realize a remarkably low OER overpotential of only η = 0.46 V, much lower than the oxygen site involving three Ni atoms in pristine β-NiOOH (η = 0.66 V) [202]. Since theoretical calculations predicted that Co, Rh, and Ir dopants would lead to low overpotentials to improve the OER activity of Ni-based hydroxides, an experimental confirmation on the altered OER activities for a series of metals (Mo, W, Fe, Ru, Co, Rh, Ir) doped into γ-NiOOH has been reported [203]. The in situ electrical conductivity for metal-doped γ-NiOOH correlated well with the trend in enhanced OER activities. The DFT calculations, which suggested that the intrinsic connections to the double exchange interaction between adjacent metal ions with various d orbital occupancies, rationalized the experimental results, serving as an indicator for the key metal-oxo radical character [203].

6.7.3. FeOx

Recent advancement and progress in initializing Fe-based OER electrocatalysts with different supporting materials, including carbon-based materials, layered double hydroxides, Prussian blue analogous, metal–organic frameworks, were reviewed by Xiong et al. [204]. In the review, the OER mechanism and some typical OER electrochemical parameters of Fe-based electrocatalysts supported on various supporting materials from experimental and theoretical viewpoints were highlighted. Some challenges and expectations for promoting the catalytic performance were described [204].

In photoelectrochemical (PEC) water oxidation on hematite (α-Fe₂O₃), the mechanism of the subsequent rate-limiting O–O bond formation step was investigated by rate law analysis based on EIS measurements and probing the reaction intermediates with operando FTIR spectroscopy. Distinct reaction orders of ~1 and ~2 were observed in near-neutral and highly alkaline environments, respectively. The unity rate law in near-neutral pH regions suggests a mechanism of WNA to –Fe=O to form the O–O bond. Operando observation of a surface superoxide species by FTIR further confirmed this pathway. In highly alkaline regions, coupling of adjacent surface trapped holes (I2M) becomes the dominant mechanism. While both are operable at intermediate pHs, the mechanism switch from I2M to WNA induced by local pH decrease was observed at a high photocurrent level as shown in Figure 14 [87]. In the recent report, transient photocurrent measurements for hematite photoanodes revealed that the OER rate has a third-order dependence on the surface hole density. A mechanism wherein the reaction proceeds by accumulating oxidizing equivalents through a sequence of one-electron oxidations of surface hydroxy groups was proposed. The key O–O bond formation step occurs by the dissociative chemisorption of a hydroxide ion involving three oxyl sites [205].

Polycrystalline γ-FeO(OH), synthesized at room temperature, was used as a stable, although reactive, anode for OER, and electrokinetic studies were performed to unravel the OER pathway [206]. The cell temperature, hydroxyl ion concentration, and the cation of the supporting electrolyte were varied, and the influence of external bias on the OER activity was recorded. Tafel slope and charge-transfer resistance values at high temperatures up to 65 °C unambiguously highlight the influence of the thermodynamic barrier and electron transfer kinetics. The faster OER kinetics on polycrystalline γ-FeO(OH) can also be attributed to an appreciably low activation energy, where the variation of the electrolyte concentration indicated a first-order dependence on the OH⁻ concentration. The deuterium isotope effect implicated the dissociation of hydroxyl ions on the polycrystalline γ-FeO(OH) as the rate-determining step. The direct effect of cations such as Li, Na, and K of the electrolyte on OER indicated a weak interaction of the cations with the surface-active [Fe^III-OH] species [206].

Fe₃O₄ with oxygen vacancies (Fe₃O₄-V_O) was synthesized via the Ar ion irradiation method and its OER activity was greatly improved by properly modulating the electron density around Fe atoms, which were evaluated with XANES and EXAFS methods. DFT results indicated the enhancement in desorption of the *OOH groups, which significantly reduced the OER reaction barrier. Actually, the Fe₃O₄-Vo catalyst showed a better overpotential than commercial RuO₂ [207].

Ni-, Co-, and Yb-doped FeO(OH) nanorod arrays grown directly on carbon cloth (CC) are synthesized by a simple one-step hydrothermal method. The doped Ni²⁺ and Co²⁺ can occupy Fe²⁺ and Fe³⁺ sites in FeO(OH), increasing the concentration of oxygen vacancies and the doped Yb³⁺ with a larger ionic radius can occupy the interstitial sites, which leads to more edge dislocations. The oxygen vacancies and edge dislocations greatly enrich the active sites in FeO(OH)/CC. In addition, DFT calculations confirmed that doping Ni²⁺, Co^2+, and Yb³⁺ modulates the electronic structure of the main active Fe sites, bringing its d-band center closer to the Fermi level and reducing the Gibbs free energy change in the rate-determining step of the OER [208].

6.7.4. MnOx

Nature uses a Mn cluster for water oxidation in PS II, and thus, water oxidation using Mn clusters is interesting in artificial water-splitting systems. An ultra-thin manganese oxide (MnO_X) was selected as a co-catalyst to modify the surface of BiVO₄ photoanode by a spray pyrolysis method [209]. The PEC measurements demonstrated that the surface charge transport efficiency strikingly increased by the MnO_X modification. After applying Ar plasma to the BiVO₄/MnOx sample, the transport efficiency further increased, and it was around 7 times higher compared with that of pristine BiVO₄ samples. The remarkable PEC performance could be attributed to the increased charge carrier density, extended carrier lifetime, and additional exposed Mn active sites on the BiVO₄ surface [209].

An α-Mn₂O₃/FTO electrocatalyst was used in nonaqueous (CH₃CN and DMF) and aqueous 0.1 M KPi (pH 7.0) solutions for kinetic studies of heterogeneous water oxidation. The rate of water oxidation was first order in catalyst concentration and in H₂O concentration. The square wave and cyclic voltammetry measurements revealed the stepwise proton-coupled electron transfer oxidations of the active Mn^II–OH₂ site to Mn^III–OH and then to Mn^IV=O and finally an electron transfer oxidation of Mn^IV=O to Mn^V=O species. The Mn^V=O species undergoes a rate-limiting O atom transfer to H₂O to give a Mn^III–OOH₂ species that, in turn, undergoes further oxidations to release O₂ [61].

An Mn–K cluster was investigated for electrochemical water oxidation. By using XAS, SEM, TEM, XRD, FTIR spectroscopy, and electrochemical methods, it was revealed that conversion into nanosized Mn oxides occurred for the cluster, and the nanosized Mn oxides are the true catalyst for water oxidation [210].

The Mn₃O₄ nanocatalyst, which exhibits superb catalytic activity for water oxidation under neutral conditions, was analyzed for the complex capacitance. By the change in Mn valence between Mn^II and Mn^IV, the charge was accumulated on the catalyst surface prior to the rate-determining O–O bond-forming step. The dissipation ratio was proposed for understanding the energy balance between charge accumulation and charge consumption for chemical O–O bond formation [211]. In Mn₃O₄ nanoparticles, a profile imaging technique was exploited to understand the correlation between surface atomic structures and the OER. The surface structures of Mn₃O₄ nanoparticles were changed by the reaction and the surface Mn ions were reconstructed. The commonly considered active sites disappeared from the reconstructed planes, whereas Mn ions were still exposed at the edges of nanoparticles. Thus, the surface reconstructions can deactivate low-index surfaces of Mn oxides in the OER process, which was further validated by DFT calculations [212].

An Mn^VII=O intermediate during electrocatalytic water oxidation by a c-disordered δ-MnOx was identified as an onset-potential-dependent reduction peak at 0.93 V. This intermediate is proven to be highly reactive and much more oxidative than permanganate ion. Thus, a new catalytic mechanism for water oxidation catalyzed by Mn oxides was proposed with the involvement of the Mn^VII=O intermediate in a resting state and the Mn^IV-O-Mn^VII=O as a real active species for O-O bond formation. Figure 15 shows the proposed catalytic cycle, involving Mn^VII=O, in the MnOx-catalyzed water-oxidation reaction. The overall mechanistic process involves charge accumulation (S₀/S₃), charge rearrangement (S₃/S₄), active-state formation (S₄/S_4′), and oxygen evolution (S_4′/S₀) [68].

6.7.5. Mixed Metal Oxides

Two or three transition metals are mixed to form oxides of high electrocatalytic performance for water electrolyzers at a low cost. An NiFe oxide catalyst was employed as the anode catalyst with an NiMo oxide cathode catalyst with a high-performance perovskite-Si tandem solar cell, achieving a record 20% STH efficiency [213]. Nickel ferrite, NiFe₂O₄, cobalt ferrite, and CoFe₂O₄, are efficient and promising anode catalyst materials in the field of electrochemical water splitting.

In Ni-Fe water oxidation electrocatalysts, Ni is likely not an active site for water oxidation, because Ni cannot achieve high-oxide states in aqueous environments at relevant potentials [214]. For the OER of NiFeO_xH_y, the addition of Co²⁺ cation increased the current density by 32.7% by the cation transport effect [215]. Using operando XAS, it was revealed that Ni oxidizes from the initial +2 oxidation state to the +3/+4 state [216]. For Ni-Fe oxyhydroxides, in situ monitoring of the Fe active site number and turn-frequency number provided important insights into the activity degradation/regeneration caused by Fe dissolution/adsorption as well as site-dependent activity and stability [217]. In the case of NiFe₂O₄, an Fe-site-assisted LOM pathway as the preferred OER mechanism was predicted. On the other hand, in the case of CoFe₂O₄, an Fe-site-assisted LOM pathway and a Co-site-assisted AEM pathway could both play a role [218].

Amorphous/crystalline NiFe₂O₄ induced by vanadium doping showed a superior electrocatalyst and long-term stability [219]. For amorphous Ni-Fe mixed metal oxides, analysis of the XAS revealed local structural transitions. A dual-site OER reaction mechanism was proposed, in which potential and rate-determining steps occur at Ni and Fe sites, respectively [220].

An Fe/Ca-based bimetallic oxide, CaFe₂O₄, exhibited outstanding OER activity in alkaline media. DFT calculations suggested an unconventional mechanism via the direct formation of O–O bonds between two oxygen intermediates, which are adsorbed on a multi-iron site on the catalyst surface [221].

On spinel NiCo₂O₄ abundant Co defects were preferentially produced by tuning the M–O bond length. Theoretical calculations and experiments proved that Al doping elongated the Co–O bond and promoted the ionization of Co under plasma treatment [222]. Spinel Co₂MnO₄ showed higher OER activity, most probably due to the ideal binding energies of the OER intermediates [223].

For modulated NiFeX and FeCoX (X = W, Mo, Nb, Ta, Re, and MoW) oxyhydroxide catalysts, in situ and ex situ soft and hard XAS were used to characterize the oxidation transition and facilitate the lower OER overpotential [224]. (Co–Fe–Pb)Ox in acidic solutions through a cobalt-selective self-healing mechanism was investigated. The kinetics of the process were investigated by soft XAS and it was revealed that low concentrations of Co²⁺ in the solution stabilize the catalytically active Co(Fe) sites [225].

6.8. Layered Double Hydroxide (LDH)

LDH are emerging catalyst materials with inner layer water molecules and higher anion exchange capacity. They have been extensively used as electrocatalytic materials owing to their high specific surface area, environmental friendliness, lower cost, and non-toxicity [226]. A kind of LDH itself may become photocatalysts for water splitting. The electronic properties, such as band structure, bandgap energy (Eg), density of states (DOS), and band edge placement for M^IIM^III-LDHs (MI^II = Mg, Co, Ni and Zn; M^III = Al and Ga) were calculated by using the DFT + U method. The band structures of Mg- and Zn-based LDHs and Co- and Ni-based LDHs are responsive to ultraviolet (Eg > 3.1 eV) and visible light (Eg < 3.1 eV), respectively. The DOS calculations revealed that the photogenerated hole localizes on the surface hydroxyl group of LDHs, facilitating the oxidization of a water molecule without a long transportation route. The band edge placements of NiGa-, CoAl-, ZnAl-, and NiAl-LDHs have a driving force (0.965 eV, 0.836 eV, 0.667 eV, and 0.426 eV, respectively), toward oxygen evolution. In the experimental observations, only CoAl-LDH was an efficient oxygen evolution photocatalyst, agreeing well with the theoretical prediction [227].

For NiFe-LDH and Ni-LDH, the critical role of superficial oxygen vacancies in enhancing electronic transport was discussed based on the electrochemical analysis by correlating with electrocatalytic activities [228]. The in situ conversion process to yield a monolayer of Ni(OH)₂ on electrodes was presented and the dynamic active site of the monolayer promoted the OER process. Doping with Co caused the oscillation of Ni and Co valence states in NiCo hydroxide. This study defined an in situ conversion process to yield monolayer LDH and a fundamental understanding of the origin of the active sites in monolayer LDHs for the OER [229]. Direct spectroscopic evidence for the different active sites in Fe-free and Fe-containing Ni oxides was reported for ultrathin LDH samples. ¹⁸O-labeling experiments in combination with in situ Raman spectroscopy were employed to probe the role of lattice oxygen as well as an active oxygen species, NiOO⁻, in the catalysts. It was found that lattice oxygen is involved in the OER for Ni and NiCo-LDHs, but not for NiFe and NiCoFe-LDHs. Moreover, NiOO⁻ is a precursor to oxygen for Ni and NiCo-LDHs, but not for NiFe and NiCoFe-LDHs [78]. For M-doped Ni-based LDH (M = Ni, Co, and Fe), the OER mechanism was investigated theoretically for the reaction processes of AEM, LOM, and IMOC (= intramolecular oxygen coupling) mechanism. Theory predicted overpotential and Tafel slopes, and the findings were in agreement with the observation. As a result, depending on the applied potential, the reaction mechanism changed [230]. In addition to electrocatalysts, NiFe-LDH may be used as the flexible electrode of Zinc-Air batteries [231].

For the Cu-NiFe-LDH electrocatalyst, a novel magnetic Fe^III site spin-splitting strategy was suggested [232]. The electronic structure and spin states of the Fe^IIII sites are effectively induced by the Jahn–Teller effect of Cu²⁺. The theoretical calculations and operando ATR FTIR revealed that the facilitation for the O–O bond formation accelerated the production of O from OH and improved the OER activity [232]. For as-prepared sulfated Co-NiFe-LDH nanosheets, the kinetic energy barrier of the O–O coupling is significantly reduced. The formation of M-OOH on the active site at low overpotential was directly confirmed in 1 M KOH solution by in situ Raman and charge transfer fitting results. In a weakly alkaline environment of 0.1 M KOH, a sequential proton–electron transfer mechanism replaces the concerted proton–electron transfer mechanism, and the proton transfer step becomes the rate-determining step (RDS) as illustrated in Figure 16 [233].

A Pt-induced NiFe-LDH (Pt-NiFe LDH) nanosheet was synthesized and large current density electrodes could be achieved in OER as well as HER [234]. At NiO/NiFe-LDH, the adsorption energy of *OH and *OOH can be adjusted independently, so as to bypass the scaling relationship and achieve high catalytic performance [235]. ZnO nanoparticles are uniformly distributed on the NiFe-LDH nanoflowers, which are prepared uniformly on the three-dimensional porous Ni foam. The active sites changed from Fe cations to Ni cations during OER and the OER dynamics were significantly improved [236]. The hierarchical bimetal nitride/hydroxide (NiMoN/NiFe LDH) array exhibits the industrially required current density. In situ electrochemical spectroscopy reveals that a hetero interface facilitates dynamic structure evolution to optimize the electronic structure. Operando EIS measurement implied the accelerated OER kinetics and intermediate evolution due to fast charge transport. For the OER mechanism, the combination of theoretical and experimental studies revealed that as-activated NiMoN/NiFe-LDH follows an LOM process with accelerated kinetics [237].

NiFe LDH@Ni₃S₂ heterostructure as an efficient bifunctional electrocatalyst for overall water splitting was prepared. Three-dimensional porous heterostructure arrays caused good electrocatalytic activity with a low Tafel slope [238]. By incorporating a semiconductor CdS/CdSe-MoS₂ and NiFe-LDH for the OER, the as-prepared photoelectrode required a potential lower than the theoretical water splitting potential. Operando XAS measurements revealed that the formation of highly oxidized Ni species under illumination provides large photocurrent gains [239]. Hetero structures of LDH with graphitic carbon nitride (g-C₃N₄) stand as promising photo- and electro-catalysts for water oxidation and reduction. Mechanisms involved in electrocatalytic, photocatalytic, and photoelectrocatalytic water splitting processes were reviewed with the necessary insights on the material [240]. By taking CuTi-LDH@g-C₃N₄ and Bi₂O₂CO₃/NiFe-LDH@g-C₃N₄ as examples, the importance of heterojunctions and interfacial chemistry in the water-splitting mechanism was explained in detail [240].

6.9. Metal–Organic Framework (MOF)

MOF-derived materials have been demonstrated with high surface area, porous structure, increased electron transport, accessible active sites, and tailorable properties. And then they may provide a new route for designing the catalysts with excellent electrochemical water oxidation activity [241]. Singh et al. discussed the MOF-derived electrocatalysts by electronic structure modulation since the recent studies of MOF-derived electrocatalysts mainly focused on morphological development, crystal structure modulation, facet engineering, and enhancement of the electrochemical surface area [241]. As the controlling factors for efficient OER, discussed were e_g orbital filling, metal–oxygen covalency, mixed valency of metal ions, octahedral vs. tetrahedral occupancy of the metal ions, and vacancy engineering. To improve the OER activity and stability of the MOF catalysts, tuning the electronic properties by interfacial modulation, surface overlayer, Fermi level manipulation, self-supported strategy, and heterostructure formation have been addressed [241].

FeNi–tannic acid coordination crystal was in situ grown on Ni foam ((FeNi)–Tan/NF) to directly catalyze the OER, and it exhibited predominant electrocatalytic OER activity [242]. A series of MOFs; FeM-MOF (M = Fe, Co, Ni, Zn, Mn; H₄L = 3,3,5,5′-azoxybenzenetetracarboxylic acid) were synthesized under a simple and mild condition. Among them, the FeCo-MOF catalyst exhibits an extremely low overpotential and small Tafel slope in an alkaline electrolyte for OER, which had far exceeded the commercial catalyst IrO₂ [243].

The surface reconstruction phenomenon of MOF-based nanomaterial electrocatalysts for OER was summarized, and the effects of structural and compositional transformation on the catalytic activity were discussed because many MOF-based catalysts inevitably undergo irreversible surface reconstruction during the redox process [244]. The causes and conditions of surface reconstruction and its influence on OER performance were also discussed, demonstrating the structure–activity relationship between surface reconstruction and catalytic performance [244]. A facile impregnation method through an ion-exchange process to fabricate Fe-doped Co-BDC nanosheets (Fe@Co-BDC NSs, BDC = benzenedicarboxylic) was proposed, though most of the related reports focus on the hydrothermal methods to prepare the mixed-metal MOFs. For an efficient OER electrocatalyst, the morphology change and the electronic structure of Fe@Co-BDC NSs were important to represent significant enhancement in the activity [245].

For MOF-based catalysts, in situ or operando Raman spectroscopic studies are useful to identify the adsorption sites, defect sites, structural or spin transitions, reaction centers, intermediates, and so on. Sunil et al. reviewed the current research for OER mechanisms with Raman spectroscopy in probing the structure, guest adsorption, catalytic activity, and reaction mechanisms of MOFs [246]. MOFs may become alternative OER catalysts because in situ self-reconstruction from MOFs to (oxy)hydroxides could be performed in alkaline electrolytes. Thus, Fe-doped Co-MOF nanosheets were prepared and utilized straightforwardly as OER electrocatalysts. CoFe-layered bimetallic hydroxides (CoFe-LDHs) with abundant active sites, which are generally actual active species, were obtained from in situ conversion of Co-MOF/Fe after etching by the KOH electrolyte. [247].

A new MOF with well-defined Co–Mo dual sites, HZIF-2-CoMo, (HZIF = Hybrid zeolitic imidazolate framework), was reported, which can promote the OER process through an unconventional Mo/Co dual-site relay mechanism. Theoretical calculations suggested that the Mo and Co sites stabilize the *OH and *OOH intermediates, respectively, and that the unique Co–O–Mo configuration induces the formation of a Co–O*–Mo transition intermediate, remarkably reducing the reaction-free energy. As a result, HZIF-2-CoMo showed one of the best OER electrocatalysts reported [248].

Two-dimensional cobalt ion (Co²⁺) and benzimidazole (bIm)-based zeolite imidazole framework nanosheets were reported as exceptionally efficient electrocatalysts for the OER. Liquid-phase ultrasonication was applied to exfoliate a Co₄(bIm)₁₆ zeolite imidazole framework (ZIF) into nanoscale sheets. The electrochemical and physicochemical characterization data supported that the OER active site is assigned to nitrogen-coordinated cobalt oxyhydroxide, N₄CoO(OH), of the exfoliated nanosheet, following a mechanism known for Co-porphyrin and related systems as shown in Figure 17 [249].

6.10. Metal Complexes

Since the CaMn₄O₅ cluster in PS II can catalyze the OER with a very low overpotential, it is expected that multinuclear water oxidation catalysts possess superior OER performances. Inspired by the CaMn₄O₅ cluster in PS II, some multinuclear complexes were synthesized that could catalyze water oxidation [250]. Mimicking the Mn₄ cluster, an artificial Mn₄ oxide cluster was proposed from the DFT calculation. The cluster may act as an OER catalyst because the ability of electron and proton transfers was theoretically exhibited [251]. Since the degradation of organic ligands is an inevitable characteristic of molecular catalysts, the ligands that pose the ability for electron transfer come under consideration [250]. By referring to the molecular mechanism of PS II, redox-active ligands were employed for metal complex catalysts and then an achievement of current efficiency of 100%, overpotential of 200–300 mV, and a turnover frequency of 100s⁻¹ has been reported [250].

Mainly with Ru, Mn, and Ir, metal complex catalysts for artificial OER were investigated at the beginning. However, recent investigations shift to non-precious natural abundant metals, such as Fe, Co, Ni, and Cu, and grow with understanding of proton-coupled electron transfer and O-O bond formation [252]. Recently, some mononuclear heterogeneous catalysts showed a high OER activity, which testified that mononuclear active sites with suitable coordination surroundings could also catalyze water oxidation efficiently. Though some mononuclear molecular complexes for OER show high water oxidation activity, it cannot be excluded that the high activity arises from the formation of dimeric species. The development of mono-/multi-nuclear homo and heterogeneous catalysts for water oxidation was focused on in a review paper [253]. This review also provided active sites and possible catalytic mechanisms of OER on the mono-/multi-nuclear catalysts [253]. Karmakar summarized the progress in research on molecular catalysts based on transition metals in the homogeneous phase while emphasizing the current mechanistic understanding of the water oxidation reaction. The factors that influence the character of water oxidation catalysts, such as reaction conditions, attached ligands, and transition metal centers, have been discussed as well [254].

6.10.1. Mn Complexes

An electrocatalytic water oxidation reaction with Mn^III tris(pentafluorophenyl) corrole (MTPC) in propylene carbonate (PC) was reported. O₂ was generated at the Mn^V/IV potential with hydroxide, but a more anodic potential was required to evolve O₂ with only water. With a synthetic Mn^V(O) complex of MTPC, a second-order rate constant was determined with hydroxide, whereas its reaction with water occurred much more slowly. Significantly, during the electrolysis of MTPC with water, an Mn^IV-peroxo species was identified with various spectroscopic methods, including UV–Vis, EPR, and FTIR spectroscopies. Isotope labeling experiments confirmed that both O atoms of this peroxo species are derived from water, suggesting the involvement of the water nucleophilic attack (WNA) mechanism in water oxidation. DFT calculations suggested that the nucleophilic attack of hydroxide on Mn^V(O) and also WNA to one electron-oxidized Mn^V(O) are feasibly involved in the catalytic cycles but that direct WNA to Mn^V(O) is not likely to be the main O–O bond formation pathway in the electrocatalytic OER by MTPC [72].

6.10.2. Fe Complexes

A pentanuclear Fe complex was reported as an efficient and robust OER catalyst with a turnover frequency of 1900 s⁻¹, which is about three orders of magnitude larger than that of other Fe-based catalysts. The electrochemical analysis confirmed the redox flexibility of the system, characterized by six different oxidation states between Fe^II₅ and Fe^III₅, where the Fe^III₅ state is active for oxidizing water. Quantum chemistry calculations indicated that the presence of adjacent active sites facilitates O–O bond formation with a low reaction barrier [255].

The kinetics of water oxidation by K₂FeO₄ have been reinvestigated by UV/Vis spectrophotometry at pH 7–9 in 0.2 M phosphate buffer. The rate of reaction was found to be second-order in both [FeO₄²⁻] and [H⁺]. These results are consistent with a proposed mechanism in which the first step involves the initial equilibrium protonation of FeO₄²⁻ to give FeO₃(OH)⁻, which then undergoes rate-limiting O–O bond formation. Analysis of the O₂ isotopic composition for the reaction in H₂¹⁸O suggests that the predominant pathway for water oxidation is intramolecular O–O coupling. DFT calculations supported the proposed mechanism [256].

6.10.3. Cu Complexes

With a family of Cu^II o-phenylene bis-oxamidate complexes, the reactivity sequence for the OER was found to be a function of the substitution pattern on the periphery of the aromatic ring. In situ EPR, FTIR, and spectroelectrochemistry suggested that ligand-centered oxidations were preferred over Cu-centered oxidations. The resonance Raman spectroelectrochemical study revealed the accumulation of a bis-imine bound Cu^II superoxide species under catalytic turnover as the reactive intermediate, which provides evidence for the O–O bond formation during the OER process [79]. Bearing the redox-active HL ligand, Cu^II(HL)(OTf)₂ (HL = N,N-bis(2,2′-bipyrid-6-yl)amine, OTf = triflate) was investigated for an OER catalyst. This Cu catalyst was found to be active as an OER catalyst at pH 11.5, at which the deprotonated complex [Cu^II(L⁻)(H₂O)]⁺ is the predominant species in the solution. The overall OER mechanism was found to be initiated by two proton-coupled electron-transfer steps. Kinetically, a first-order dependence in the catalyst and a zeroth-order dependence in the phosphate buffer were found. A computational study supported the formation of a Cu–oxyl intermediate, [Cu^II(L•)(O•)(H₂O)]⁺. From this intermediate onward, the O–O bond is formed via a single-electron transfer from an approaching hydroxide ion to the ligand. Throughout the mechanism, the Cu^II center is proposed to be redox-inactive [257].

The mechanism of OER catalyzed by a mononuclear Cu complex in alkaline conditions was studied by DFT calculations and shown in Figure 18. Firstly, a water molecule coordinating with the Cu center of the Cu^II complex generates the complex of Cu^II–H₂O. This complex undergoes two proton-coupled electron transfer processes to produce an intermediate (•L–Cu^II–O•), which can be described as a Cu^II center interacting with a ligand radical antiferromagnetically and an oxyl radical ferromagnetically. The oxidation process occurs mainly on the ligand moiety, which can trigger O–O bond formation via the WNA mechanism. The attacking water transfers one of the protons to the HPO₄²⁻ coupled with an electron transfer to the ligand radical, which generates a transient •OH interacting with the oxyl radical and H₂PO₄⁻. Then, the O–O bond is formed through the direct coupling of the oxyl radical and the OH radical. The triplet di-oxygen could be released after two oxidation processes. The O–O bond formation was suggested to be the rate-limiting step. Thus, the Cu complex catalyzes water oxidation with the help of a redox non-innocent ligand and HPO₄²⁻ [258].

6.11. Single-Atom Catalysts (SACs)

During the past few years, several research efforts have also unveiled the immense potential of SACs for the OER. Excellent OER performance has been demonstrated for both noble metal and non-noble metal SACs with lab-scale electrodes. The long-term stability of promising SACs is less well explored though, and considerable efforts are still required to assess this under technologically relevant conditions [259].

Atomically dispersed Ir atoms incorporated into spinel Co₃O₄ lattice as an acidic OER catalyst were reported to exhibit excellent activity and stability for water oxidation. FTIR observation of *OOH indicates that the AEM rather than the LOM dominates the OER process [83]. An electrocatalyst with Ru-atom-array patches supported on α-MnO₂ (Ru/MnO₂) for the OER was investigated. Ru/MnO₂ showed a high activity and outstanding stability with small degradation after 200 h operation. Operando vibrational and mass spectroscopy measurements were performed to probe the reaction intermediates and gaseous products for validating the OER pathway, suggesting a mechanism that involves only *O and *OH species as intermediates. This mechanism allows direct O–O radical coupling for O₂ evolution, i.e., OPM process in Figure 3c was suggested. First-principles calculations confirmed the cooperative catalysis mechanism with a reduced energy barrier. Time-dependent elemental analysis demonstrated the occurrence of the in situ dynamic cation exchange reaction during the OER, which is the key for triggering the reconstruction of Ru atoms into the ordered array with high durability [80].

A macromolecule-assisted SAC providing high-density Co single atoms (10.6 wt% Co) in a pyridinic N-rich graphenic network was reported [260]. The highly porous carbon network with increased conjugation and vicinal Co site decoration significantly enhanced the electrocatalytic OER in 1 M KOH with more than 300 h stability. Operando X-ray absorption near-edge structure demonstrated the formation of electron-deficient Co-O coordination intermediates, accelerating OER kinetics [260].

A representative set of 11 transition metal atoms (Sc, V, Ti, Cr, Mn, Fe, Co, Ni, Cu, Pd, Pt) anchored on nitrogen-doped graphene were considered by means of DFT calculations [261]. Most of them form stable superoxo and peroxo intermediates when they react with molecular oxygen, which has a direct impact on the OER. Thus, in the corresponding microkinetic models, this step of the reaction cannot be neglected. A representative Gibbs energy profile is shown in Figure 19. Depending on the transition metal atom, the inclusion of the superoxo/peroxo complexes in the analysis of the reaction profile can change the kinetics by several orders of magnitude [261]. To clarify the assignment of the intermediates, a DFT technique was applied to a set of 30 SACs made by ten metal atoms (Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Pd, and Pt) anchored on three widely used 2D carbon-based materials; graphene, nitrogen-doped grapheme, and carbon nitride. On SACs, it has been generally assumed that OER occurs via the formation of three intermediates, *OH, *O, and *OOH species, they could be referred to as M(OH), M(O), and M(OOH). In all cases, however, other intermediates M(OH)₂, M(O)(OH), M(O)₂, and M(O₂) were formed with higher stabilities [262].

Compared with SACs, dual-atom catalysts (DACs) are attracting more attention including higher metal loading, more versatile active sites, and excellent catalytic activity. Several general synthetic strategies and structural characterization methods of DACs were introduced, and the involved catalytic mechanisms were discussed in the review paper [263]. Dual-atom catalysts, particularly those with heteronuclear active sites, have the potential to outperform the well-established single-atom catalysts for OER. A large-scale DFT is employed to explore the feasibility of the *O-*O coupling mechanism, which can circumvent the scaling relationship by improving the catalytic performance of N-doped graphene-supported Fe-, Co-, Ni-, and Cu-containing heteronuclear dual-atom catalysts [264].

6.12. Effect of Surface Functionalization

It has been found that the addition of carbonate salts to Pt-loaded TiO₂ suspensions led to highly efficient stoichiometric photocatalytic decomposition of liquid water into H₂ and O₂. Since a high concentration of carbonate ions is essential for the catalytic photodecomposition of water, it was considered that the carbonate species aids the desorption of O₂ from the TiO₂ surface [265]. To elucidate the effect of carbonates, DFT calculations are performed to study the photoinduced H₂O and H₂CO₃ oxidation mechanisms on TiO₂ and BiVO₄ [266]. The computational results verify that the adsorbed H₂CO₃ molecule is easily photo-oxidized compared with the adsorbed H₂O molecule, facilitating the formation of the peroxide intermediate and improving O₂ evolution and H₂O₂ production [266].

The use of surface functionalization with phosphate ion groups (Pi) enhances the interfacial proton transfer. As the results, the Pi functionalization on La₀.₅Sr₀.₅CoO_3−δ and LaCoO₃ gave rise to a significant enhancement of the OER activity when compared to La₀.₅Sr₀.₅CoO_3−δ and LaCoO₃, respectively. It was demonstrated that the surface functionalization by Pi enhanced the activity when the OER kinetics are limited by the proton transfer as shown in Figure 20. By Sr²⁺ cation substitution, O 2p band closed to the Fermi level, assisting the deprotonation step from –OOH to –OO•. Thus, depending on the position of the Fermi level, the function of Pi changed from keeping water near the surface to assisting the deprotonation reaction [267].

The hydrated cobalt phosphate (CoPi) co-catalyst was investigated for O–O bond and OOH formation, based on the conventional experimental findings. Theoretical calculations of hydrated CoPi cluster models elucidated the roles of phosphate as a source of oxygen and deliverer of protons, both of which result in the spontaneous formation of an O–O bond after the release of two electrons and two protons. The calculations also show that OOH formation proceeds subsequently depending on the spin electronic states of the hydrated CoPi surface, and O₂ formation then spontaneously progresses after the release of two electrons and two protons [268].

A proton acceptor, TA²⁻ (terephthalate ion) can mediate proton transfer pathways by preferentially accepting protons, which optimizes the O−H adsorption/activation process and reduces the kinetic barrier for O–O bond formation. A proton-transfer-promotion mechanism for OER electrocatalysts with FeO₆/NiO₆ units is proposed by in situ Raman spectroscopy, catalytic tests, and theoretical calculations [269].

The strong adsorption of hexadecyl trimethyl ammonium cations on the surface of electrocatalysts provides the increased absolute number of OH⁻ ions near the electrocatalyst surface, which effectively promotes the OER performance of electrocatalysts, such as Fe_1−yNi_yS₂@Fe_1−xNi_xOOH micro platelets and SrBaNi₂Fe₁₂O₂₂ powders [270].

7. Perspectives

In the brimming research articles of the OER developments, the catalysts for electrolyzers (electrocatalysts) are dominant because they are actually working in industries. These catalysts for electrolyzers may not be directly applied for artificial photosynthesis using photoelectrodes or photocatalysts. However, the molecular mechanism is surely applicable under the consideration of difference in the reaction conditions. Since the electrochemical OER catalysts are used with electrolytes, they usually work in alkaline or acid solution. However, photoelectrodes and photocatalysts are usually used in moderate pH values. Then, the most attention should be paid to the pH effects on the reaction mechanism when electrocatalysts are consulted for photocatalysts.

Nature assembled PS II with CaMn₄O₅ clusters as the ultimate OER catalysts. Compared with the present artificial catalysts, the molecular mechanism of the PS II catalytic reaction is different in the following issues. (i) Immediately after the O-O bond formation O₂ is released from the cluster, while the lifetime of the *OO⁻ or *OO* in artificial catalysts is so long to be easily detected. Therefore, the catalysts, of which O-O intermediate species can be detected, may not be really generalized as a good catalyst. (ii) In the CaMn₄O₅ cluster, the four transition metals share each positive hole to give four Mn^IV ions. In the case of artificial OER catalysts, one or two metal ions participate the four-electron reaction. Therefore, intermediate species of the OER should receive the next hole at the same metal ions. Namely, the catalysts of AEM process may have a difficulty to improve the activity. Then, catalysts with more than one metal atom appear promising in the future. (iii) The pathways of proton and water molecules are controlled in the PS II process. To keep the pathways, the surroundings of the Mn₄ cluster is rather hydrophobic. The artificial OER catalysts are usually surrounded by water molecules and the whole surface of the catalysts can adsorb water molecules. To improve the OER activity of the artificial catalysts, controlling the proton transfers may become important, and it could be achieved by the surface functionalization of the present catalysts. The discussion in Section 6.12 may provide a clue to the improvement of the OER activity.